通过机器学习方法研究青少年抑郁

Background: Adolescent depression is highly prevalent in low- and middle-income countries (LMICs). Identifying top key risk factors is necessary to inform effective prevention program design. Machine learning (ML) offers a powerful approach to analyze complex multidomain of data to identify the most relevant predictors and estimate risks for mental health problems. This paper applies ML to study risks for adolescent depression to enhance adolescent depression prevention efforts in LMICs. Methods: Six ML approaches (e.g., Explainable Boosting Machine, random forests, and XGBoost) were applied to study the risks of depression. Data were drawn from a digital health intervention study conducted in Kenya (year 2024–2025, n = 269). Multiple domains of childhood and adolescent adversity and stress experiences were used to predict adolescent depression (using PHQ9-A). Findings: We found that ML was a valuable approach in the early identification of adolescents at risk for depression. Among the six ML approaches applied, the random forest approach outperformed other ML approaches, especially when multiple domains of risks were included. We also found that childhood adversity or home adversity alone were not strong predictors for depression. Adding adolescent stress experiences and community school adversity experiences significantly improves the accuracy and predictability of depression. Using the top-15 and top-20 ranking factors, we achieved 74.8% and 75.1% accuracy in depression prediction, which was similar to the accuracy when all 49 adverse/stress factors were included in the predictive model (78.3%). Conclusions: Innovative ML and modern predictive modeling approaches have the potential to transform modern preventive mental health care by better utilizing multidomain data to identify individuals at risk for developing depression early and identify top risk factors (for targeted individuals and/or populations). Findings from ML can inform tailored intervention design to better mitigate risks in order to prevent depression problem development. They can also inform the better utilization of resources to target high-need cases and key determinants, which is particularly useful for LMICs and low-resource settings. This paper illustrates an example of how to move toward this direction. Future research is needed to validate the approach.

Background Major depressive disorder (MDD) in adolescents poses an increasing global health concern, yet current screening practices rely heavily on subjective reports. Virtual reality (VR), integrated with multimodal physiological sensing (EEG+ET+HRV), offers a promising pathway for more objective diagnostics. Methods In this case-control study, 51 adolescents diagnosed with first-episode MDD and 64 healthy controls participated in a 10-minute VR-based emotional task. Electroencephalography (EEG), eye-tracking (ET), and heart rate variability (HRV) data were collected in real-time. Key physiological differences were identified via statistical analysis, and a support vector machine (SVM) model was trained to classify MDD status based on selected features. Results Adolescents with MDD showed significantly higher EEG theta/beta ratios, reduced saccade counts, longer fixation durations, and elevated HRV LF/HF ratios (all p <.05). The theta/beta and LF/HF ratios were both significantly associated with depression severity. The SVM model achieved 81.7% classification accuracy with an AUC of 0.921. Conclusions The proposed VR-based multimodal system identified robust physiological biomarkers associated with adolescent MDD and demonstrated strong diagnostic performance. These findings support the utility of immersive, sensor-integrated platforms in early mental health screening and intervention. Future work may explore integrating the proposed multimodal system into wearable or mobile platforms for scalable, real-world mental health screening.

No abstract available

Depression is a major cause of disability and mortality for young people worldwide and is typically first diagnosed during adolescence. In this work, we present a machine learning framework to predict adolescent depression occurring between ages 12 and 18 years using environmental, biological, and lifestyle features of the child, mother, and partner from the child’s prenatal period to age 10 years using data from 8467 participants enrolled in the Avon Longitudinal Study of Parents and Children (ALSPAC). We trained and compared several cross-sectional and longitudinal machine learning techniques and found the resulting models predicted adolescent depression with recall (0.59 ± 0.20), specificity (0.61 ± 0.17), and accuracy (0.64 ± 0.13), using on average 39 out of the 885 total features (4.4%) included in the models. The leading informative features in our predictive models of adolescent depression were female sex, parental depression and anxiety, and exposure to stressful events or environments. This work demonstrates how using a broad array of evidence-driven predictors from early in life can inform the development of preventative decision support tools to assist in the early detection of risk for mental illness.

Abstract Background The prevalence of adolescent mental health conditions such as depression and anxiety has significantly increased. Despite the potential of machine learning (ML), there is a shortage of models that use real-world data (RWD) to enhance early detection and intervention for these conditions. Objective This study aimed to identify depression and anxiety in adolescents using ML techniques on RWD and social determinants of health (SDoH). Methods We analyzed RWD of adolescents aged 10‐17 years, considering various factors such as demographics, prior diagnoses, prescribed medications, medical procedures, and laboratory measurements recorded before the onset of anxiety or depression. Clinical data were linked with SDoH at the block-level. Three separate models were developed to predict anxiety, depression, and both conditions. Our ML model of choice was Extreme Gradient Boosting (XGBoost) and we evaluated its performance using the nested cross-validation technique. To interpret the model predictions, we used the Shapley additive explanation method. Results Our cohort included 52,054 adolescents, identifying 12,572 with anxiety, 7812 with depression, and 14,019 with either condition. The models achieved area under the curve values of 0.80 for anxiety, 0.81 for depression, and 0.78 for both combined. Excluding SDoH data had a minimal impact on model performance. Shapley additive explanation analysis identified gender, race, educational attainment, and various medical factors as key predictors of anxiety and depression. Conclusions This study highlights the potential of ML in early identification of depression and anxiety in adolescents using RWD. By leveraging RWD, health care providers may more precisely identify at-risk adolescents and intervene earlier, potentially leading to improved mental health outcomes.

No abstract available

Abstract Background This study leveraged machine learning to evaluate the contribution of information from multiple developmental stages to prospective prediction of depression and anxiety in mid-adolescence. Methods A community sample (N = 374; 53.5% male) of children and their families completed tri-annual assessments across ages 3–15. The feature set included several important risk factors spanning psychopathology, temperament/personality, family environment, life stress, interpersonal relationships, neurocognitive, hormonal, and neural functioning, and parental psychopathology and personality. We used canonical correlation analysis (CCA) to reduce the large feature set to a lower dimensional space while preserving the longitudinal structure of the data. Ablation analysis was conducted to evaluate the relative contributions to prediction of information gathered at different developmental periods and relative to previous disorder status (i.e. age 12 depression or anxiety) and demographics (sex, race, ethnicity). Results CCA components from individual waves predicted age 15 disorder status better than chance across ages 3, 6, 9, and 12 for anxiety and 9 and 12 for depression. Only the components from age 12 for depression, and ages 9 and 12 for anxiety, improved prediction over prior disorder status and demographics. Conclusions These findings suggest that screening for risk of adolescent depression can be successful as early as age 9, while screening for risk of adolescent anxiety can be successful as early as age 3. Assessing additional risk factors at age 12 for depression, and going back to age 9 for anxiety, can improve screening for risk at age 15 beyond knowing standard demographics and disorder history.

No abstract available

Background Machine learning based clinical decision support systems (CDSSs) have been proposed as a means of advancing personalized treatment planning for disorders, such as depression, that have a multifaceted etiology, course, and symptom profile. However, machine learning based models for treatment selection are rare in the field of psychiatry. They have also not yet been translated for use in clinical practice. Understanding key stakeholder attitudes toward machine learning based CDSSs is critical for developing plans for their implementation that promote uptake by both providers and families. Methods In Study 1, a prototype machine learning based Clinical Decision Support System for Youth Depression (CDSS-YD) was demonstrated to focus groups of adolescents with a diagnosis of depression (n = 9), parents (n = 11), and behavioral health providers (n = 8). Qualitative analysis was used to assess their attitudes towards the CDSS-YD. In Study 2, behavioral health providers were trained in the use of the CDSS-YD and they utilized the CDSS-YD in a clinical encounter with 6 adolescents and their parents as part of their treatment planning discussion. Following the appointment, providers, parents, and adolescents completed a survey about their attitudes regarding the use of the CDSS-YD. Results All stakeholder groups viewed the CDSS-YD as an easy to understand and useful tool for making personalized treatment decisions, and families and providers were able to successfully use the CDSS-YD in clinical encounters. Parents and adolescents viewed their providers as having a critical role in the use the CDSS-YD, and this had implications for the perceived trustworthiness of the CDSS-YD. Providers reported that clinic productivity metrics would be the primary barrier to CDSS-YD implementation, with the creation of protected time for training, preparation, and use as a key facilitator. Conclusions Machine learning based CDSSs, if proven effective, have the potential to be widely accepted tools for personalized treatment planning. Successful implementation will require addressing the system-level barrier of having sufficient time and energy to integrate it into practice.

Background: Adolescents are particularly vulnerable to mental disorders, with over 75% of cases manifesting before the age of 25. Research indicates that only 18 to 34% of young people experiencing high levels of depression or anxiety symptoms seek support. Digital tools leveraging smartphones offer scalable and early intervention opportunities. Objective: Using a novel machine learning framework, this study evaluated the feasibility of integrating active and passive smartphone data to predict mental disorders in non-clinical adolescents. Specifically, we investigated the utility of the Mindcraft app in predicting risks for internalising and externalising disorders, eating disorders, insomnia and suicidal ideation. Methods: Participants (N=103; mean age 16.1 years) were recruited from three London schools. Participants completed the Strengths and Difficulties Questionnaire, the Eating Disorders-15 Questionnaire, Sleep Condition Indicator Questionnaire and indicated the presence/absence of suicidal ideation. They used the Mindcraft app for 14 days, contributing active data via self-reports and passive data from smartphone sensors. A contrastive pretraining phase was applied to enhance user-specific feature stability, followed by supervised fine-tuning. The model evaluation employed leave-one-subject-out cross-validation using balanced accuracy as the primary metric. Results: The integration of active and passive data achieved superior performance compared to individual data sources, with mean balanced accuracies of 0.71 for SDQ-High risk, 0.67 for insomnia, 0.77 for suicidal ideation and 0.70 for eating disorders. The contrastive learning framework stabilised daily behavioural representations, enhancing predictive robustness. This study demonstrates the potential of integrating active and passive smartphone data with advanced machine-learning techniques for predicting mental health risks.

No abstract available

No abstract available

Children face trauma due to numerous factors such as parental fights, bullied in school, being below average student, racist comments, physical abuse etc. This trauma deteriorates the child's mental health and may lead to depression as well. Medical field is immensely dedicated to combat the depression among children and one of such approaches adapted includes the use Machine Learning models to predict child depression. This research work is focusing on seven different machine learning algorithms, including SVM Poly, K-Nearest Neighbor, SVM RBF, SVM Linear, Decision Tree, Gradient Boosting, Random Forest and Logistic Regression. This research examines the performance of these algorithms in detecting depressive symptoms in a child and adolescent population. In accordance with the outcomes of the research, SVM Linear algorithm emerged as the most accurate method, with an excellent accuracy rate of 95.54%. This algorithm demonstrated higher sensitivity in detecting depression symptoms in children and teens opening the possibility to prompt intervention and assistance.

This study examines the impact of academic demands on depressive symptoms among high school students. To this end, we collected a sample dataset of 1303 questionnaires modeling the relationship between academic demands and depressive symptoms among high school students. By applying a random forest model, we performed an in-depth predictive analysis on the data. To better understand the main structure and characteristics of the data, we also performed principal component analysis (PCA). Preliminary results showed significant relationships between academic stress, test scores and family support and depressive symptoms. This study not only provides a new perspective for understanding the impact of academic demands on adolescent mental health, but also demonstrates the application potential of machine learning and statistical methods in psychological research.

BACKGROUND Machine learning (ML) algorithms based on various clinicodemographic, psychometric, and biographic factors have been used to predict depression, suicidal ideation, and suicide attempt in adolescents, but there is still a need for more accurate and efficient models for screening the general adolescent population. In this study, we compared various ML methods to identify a model that most accurately predicts suicidal ideation and level of depression in a large cohort of school-aged adolescents. METHODS Ten psychological scale scores and 20 sociodemographic parameters were collected from 10,243 Chinese adolescents in the first or second year of middle school and high school. These variables were then included in a random forest (RF) model, support vector machine (SVM) model, and decision tree model for factor screening, dichotomous prediction of suicidal ideation (yes/no), and trichotomous prediction of depression (no depression, mild-moderate depression, or major depression). RESULTS The RF model demonstrated greater accuracy for predicting suicidal ideation (mean accuracy (ACC) = 87.3 %, SD = 3.2 %, area under curve (AUC) = 92.4 %) and depressive status (ACC = 84.0 %, SD = 2.8 %, AUC = 90.1 %) than SVM and decision tree models. We have also used the RF model to predict adolescents with both depression and suicidal ideation with satisfactory results. Significant differences were found in several sociodemographic parameters and scale scores among classification groups and differences in six factors between sexes. CONCLUSIONS This RF model may prove valuable for predicting suicidal ideation, depression, and non-suicidal self-injury among the general population of Chinese adolescents.

No abstract available

This study investigates the current mental health status among children and adolescents in Jiangsu Province by analyzing symptoms of depression, anxiety, and stress using standardized psychological scales. Machine learning models were utilized to identify key influencing variables and predict mental health outcomes, aiming to establish a rapid psychological well-being assessment framework for this population. A cross-sectional survey was conducted via random cluster sampling across 98 counties (cities/districts) in Jiangsu Province, enrolling 141,725 students (47,502 primary, 47,274 junior high, 11,619 vocational high school students, and 35,330 senior high ). The study focused on prevalent mental health disorders and associated risk factors. Depression, anxiety, and stress scores served as dependent variables, with 57 socio-demographic and behavioral factors as independent variables. Five supervised machine learning models (Decision Tree, Naive Bayes, Random Forest, K-Nearest Neighbors (KNN), and XGBoost) were implemented using R software. Model performance was evaluated using accuracy, precision, recall, F1 Score and Area Under the ROC Curve (AUC). Feature importance analysis was conducted to identify key predictors. The study revealed significant mental health disparities: depression (14.9%), anxiety (25.5%), and stress (10.9%) prevalences showed clear gender and regional gradients. Females exhibited higher rates across all conditions (p < 0.05), and urban areas had elevated risks compared to suburban regions. Mental health deterioration escalated with educational stages (e.g., depression from 9.2% in primary to 21.2% in senior high; χ²trend = 2274.55, p < 0.05). The XGBoost model demonstrated optimal predictive performance (AUC: depression = 0.799, anxiety = 0.770, stress = 0.762), outperforming other models. Feature importance analysis consistently identified bullying duration, age, and drinking history as top risk factors across both Gain and SHAP methods, while SHAP values additionally emphasized modifiable lifestyle factors (e.g., breakfast frequency) and demographic variables (e.g., gender). This study identifies bullying, age, and alcohol consumption history as key mental health risk factors among Jiangsu’s children and adolescents. These findings emphasize the need for school-based anti-bullying programs, age-specific mental health counseling, and healthy lifestyle education (including alcohol refusal). Lifestyle behaviors like daily breakfast intake should be integrated into dietary interventions for mental health promotion. Urban-rural and gender disparities necessitate targeted support for urban adolescent females, while educational stage differences highlight the criticality of early prevention.

Background: Current assessments of adolescent emotional and behavioral problems rely heavily on subjective reports, which are prone to biases. Aim: This study is the first to explore the potential of speech signals as objective markers for predicting emotional and behavioral problems (hyperactivity, emotional symptoms, conduct problems, and peer problems) in adolescents using machine learning techniques. Materials and Methods: We analyzed speech data from 8215 adolescents aged 12–18 years, extracting four categories of speech features: mel-frequency cepstral coefficients (MFCC), mel energy spectrum (MELS), prosodic features (PROS), and formant features (FORM). Machine learning models—logistic regression (LR), support vector machine (SVM), and gradient boosting decision trees (GBDT)—were employed to classify hyperactivity, emotional symptoms, conduct problems, and peer problems as defined by the Strengths and Difficulties Questionnaire (SDQ). Model performance was assessed using area under the curve (AUC), F1-score, and Shapley additive explanations (SHAP) values. Results: The GBDT model achieved the highest accuracy for predicting hyperactivity (AUC = 0.78) and emotional symptoms (AUC = 0.74 for males and 0.66 for females), while performance was weaker for conduct and peer problems. SHAP analysis revealed gender-specific feature importance patterns, with certain speech features being more critical for males than females. Conclusion: These findings demonstrate the feasibility of using speech features to objectively predict emotional and behavioral problems in adolescents and identify gender-specific markers. This study lays the foundation for developing speech-based assessment tools for early identification and intervention, offering an objective alternative to traditional subjective evaluation methods.

Depressive symptoms in adolescents can significantly affect their daily lives and pose risks to their future development. These symptoms may be linked to various factors experienced during both childhood and adolescence. Machine learning (ML) has attracted substantial attention in the field of adolescent depression; however, studies establishing prediction models have primarily considered childhood or adolescent features separately, resulting in a lack of analyses that incorporate factors from both stages. We collected 39 features related to childhood and adolescence. Using the maximum relevance-minimum redundancy method and four ML algorithms, we determined the optimal feature subset for identifying depressive symptoms and constructed child-adolescent models. Stepwise logistic regression and four ML methods were employed to create demographic and combined models, respectively. The performance of each model was evaluated using a test set, and SHapley Additive exPlanations (SHAP) were utilized to interpret the models’ prediction results. The proposed child-adolescent models exhibited superior performance on the test set than the demographic and combined models (AUC: 0.835–0.879 versus 0.530 and 0.840–0.876, respectively). The optimal feature subset included two childhood features (relationship quality with peers and parental absence) and four adolescence features (social trust, academic pressure, importance of the internet for entertainment, and positive parenting behaviour). These features were found to be more effective than demographic characteristics in distinguishing depressive symptoms in adolescents. This study demonstrates the correlation between adolescence depressive symptoms and specific factors from both childhood and adolescence, as well as the potential of ML to predict it. These findings may serve as a reference for future intervention studies.

The co-occurrence of depression and anxiety in adolescents is associated with a greater risk than the presence of depressive symptoms alone. Predicting anxiety disorders among depressed adolescents is critical for interventions and therapeutic tools. We recruited 2316 depressed adolescents through the Chinese Adolescent Depression Cohort (CADC) and collected 34 predictive factors for model construction. The Light Gradient Boosting Machine (LightGBM) prediction model and Shapley Additive Explanations (SHAP) algorithm were implemented for in-depth interpretation of the predictive importance of different factors. Furthermore, chi-square automatic interaction detection (CHAID) and ordinal logistic regression were used to explore the factor interactions and validate the importance of the SHAP value-based factors, respectively. Nine key risk factors were identified. In addition to depressive severity, rumination, perceived stress, sleep quality, alexithymia, peer victimization, academic stress level, emotion-focused coping, and parental overprotect were recognized as key risk factors for the onset of anxiety. Resilience was recognized as a protective factor. Interaction analysis captured critical interactions between depression and six other risk factors in relation to different levels of anxiety risks. Interactions between the protective effect of resilience and four risk factors were also analyzed. High-risk subgroups and low-risk groups for different levels of anxiety were identified through the CHAID decision tree. The high-risk subgroups for severe anxiety include (1) adolescents with severe depression symptoms, (2) with moderate depression symptoms and high rumination, and (3) with severe depression symptoms and high alexithymia. The low-risk subgroups are (1) adolescents with low depression and rumination, (2) with low depression, low alexithymia, and more parental care, (3) with low depression, moderate rumination, and moderate academic stress. Utilizing an explainable machine learning approach enables us to identify the risk and protective factors for anxiety disorders among depressed adolescents. The SHAP analysis results suggest that depression severity was the most important predictor for co-morbid anxiety. CHAID decision tree further identified risk subgroups. These findings suggest that clinical workers take into consideration the above risk and protective factors as well as their interactions to develop appropriate therapies for the prevention of comorbid anxiety with depression.

Background: As adolescent suicide rates continue to rise, innovation in risk identification is warranted. Machine learning can identify suicidal individuals based on their language samples. This feasibility pilot was conducted to explore this technology’s use in adolescent therapy sessions and assess machine learning model performance. Method: Natural language processing machine learning models to identify level of suicide risk using a smartphone app were tested in outpatient therapy sessions. Data collection included language samples, depression and suicidality standardized scale scores, and therapist impression of the client’s mental state. Previously developed models were used to predict suicidal risk. Results: 267 interviews were collected from 60 students in eight schools by ten therapists, with 29 students indicating suicide or self-harm risk. During external validation, models were trained on suicidal speech samples collected from two separate studies. We found that support vector machines (AUC: 0.75; 95% CI: 0.69–0.81) and logistic regression (AUC: 0.76; 95% CI: 0.70–0.82) lead to good discriminative ability, with an extreme gradient boosting model performing the best (AUC: 0.78; 95% CI: 0.72–0.84). Conclusion: Voice collection technology and associated procedures can be integrated into mental health therapists’ workflow. Collected language samples could be classified with good discrimination using machine learning methods.

Abstract Background. Despite the growing recognition of adolescent suicide as a pressing concern, traditional methods for identifying suicide risk often fail to capture the complex interplay of socio-ecological and psychological factors. The advent of machine learning (ML) offers a transformative opportunity to improve suicide risk prediction and intervention strategies. Objective. This study aims to utilize ML techniques to analyze socio-ecological and psychological risk factors to predict suicide ideation, plans and attempts among a nationally representative sample of Ghanaian adolescents. Methods. A cross-sectional survey was conducted with 1,703 adolescents aged 12–18 years across Ghana measuring psychological factors (depression symptoms, anxiety symptoms etc) and socio-ecological factors (bullying, parental support etc) using validated measures. Descriptive statistics were conducted and random forest and logistic regression models were employed for suicide risk prediction, i.e., ‘ideation, plans and attempts’. Model performance was evaluated using accuracy, sensitivity, specificity and feature importance analysis. Results. Psychological factors such as depression symptoms (r = .42, p < .01), anxiety (r = .38, p < .01) and perceived stress (r = .35, p < .01) were the strongest predictors of suicide ideation, plans and attempts, while parental support emerged as a significant protective factor (r = −.34, p < .01). The random forest model demonstrated good predictive performance (accuracy = 78.3%, AUC = 0.81). Gender differences were observed. Conclusions. This study is the first to apply ML techniques to a nationally representative dataset of Ghanaian adolescents for suicide risk prediction, i.e., ‘ideation, plans and attempts’. The findings highlight the potential of ML to provide precise tools for early identification of at-risk individuals.

Objectives To investigate individual, interpersonal, health system, and community factors associated with suicidal ideation (SI) and attempts (SA). Methods Utilizing nationally representative data from the National Longitudinal Study of Adolescent to Adult Health (7th-12th graders in 1994-95 followed >20 years until 2016-18, N=18,375), least absolute shrinkage selector operator (LASSO) regression determined multilevel predictors of SA and SI. Models comprised full and diagnosis subgroups (ADD/ADHD, depression, PTSD, anxiety, learning disabilities [LD]). Results Approximately 2.48% and 8.97% reported SA and SI, respectively. Over 25% had depression, and 20.98% anxiety, 6.42% PTSD, 4.55% ADD/ADHD, and 2.50% LD. LASSO regression identified 20 and 21 factors associated with SA and SI. Individual-level factors associated with SI and SA included educational attainment, substance use, ADD/ADHD, depression, anxiety, and PTSD. Interpersonal-level factors included social support, household size, and parental education, while health system-level factors comprised health care receipt, health insurance, and counseling. The strongest associations were among individual-level factors followed by interpersonal and health system factors. Conclusions The distinct factors associated with SI and SA across diagnostic subgroups highlight the importance of targeted, subgroup-specific suicide prevention interventions. These findings emphasize the value of precise, data-driven approaches for suicide prevention among diverse populations and individuals with disabilities across the life-course.

Multimodal analysis of Magnetic Resonance Imaging (MRI) data enables leveraging complementary information across multiple imaging modalities that may be incomplete when using a single modality. For brain connectivity analysis, graph-based methods, such as graph signal processing, are effective for capturing topological characteristics of the brain structure while incorporating neural activity signals. However, for tasks like group classification, these methods often rely on traditional machine learning algorithms, which may not fully exploit the underlying graph topology. Recently, Graph Convolutional Networks (GCN) have emerged as a powerful tool in brain connectivity research, uncovering complex nonlinear relationships within the data. Here, we develop a multimodal GCN model to jointly model brain structure and function to classify anxiety and depression in adolescents using the Boston Adolescent Neuroimaging of Depression and Anxiety dataset. The graph’s topology is initialized from structural connectivity derived from diffusion MRI, while functional connectivity is incorporated as node features to improve distinction between anxious, depressed patients and healthy controls. Interpretation of key brain regions contributing to classification is enabled through Gradient-weighted Class Activation Mapping, revealing the influence of the frontal and limbic lobes in the diagnosis of the conditions, which aligns with previous findings in the literature. By comparing classification results and the most discriminative features between multimodal and unimodal GCN-based approaches, we demonstrate that our framework improves accuracy in most classification tasks and reveals significant patterns of brain alterations associated with anxiety and depression.

Detecting mental health problems at the earliest can help professionals provide correct treatment and increase the value of patients. One of the most prevalent types of impairment in the world is depression. It is important to deal adolescent mental health issues which can escalate to more serious problems if not addressed early. Machine learning techniques are ideal for estimating and diagnosing medical data. Feature selection techniques have been applied to the entire attribute dataset to reduce attributes. Machine Learning accuracy is associated with attributes from the dataset. Three algorithms are employed in this work, Logistic Regression, Random Forest Classifiers, and Decision Tree Classifiers, in data science techniques. Sensitivity and Specificity are measures that estimate a model’s capability to predict true positive and negative results. When these algorithms are compared the random forest classifier yields 95% accuracy when compared with other algorithms.

Adolescent depression is more harmful because teens are in a critical period of mental development. The combination of speech analysis and machine learning techniques obtains promising results in detecting depression. However, the current research mainly focuses on enhancing the performance of models and lacks explanation of the model decision process. This paper aims to investigate which features are of importance in identifying adolescent depression and how these features affect the predictions of the model. We extracted 225 acoustic features form the recordings of teenagers reading three Mandarin paragraphs, and then built an ensenble machine learning model that achieved a mean F1 score of 0.853. Combined with the model interpretation framework SHAP, we found that the dispersion of the first formant contributed most to the predictions. Our work is transparent to the process of model decision making, which may promote the the application of machine learning in healthcare fields.

The objective of this study is to explore the application of machine learning (ML) for the prediction of relapse events in adolescents suffering from Major Depressive Disorder using actigraphy and ecological momentary assessment (EMA) data. The data were collected from 114 adolescents aged between 12 and 21 who were participating in a depression research study at the Centre for Addiction and Mental Health. They made up to 8 visits where a psychiatrist would assess their level of depression by conducting a Children's Depression Rating Scale (CDRS) survey. Between visits, participants wore the GeneActiv wearable device to provide actigraphy data and also answered survey questions to provide EMA data. Any subject that experienced one visit with a CDRS score of less than 40 followed by another visit with a score of greater than 40 was labelled as undergoing a relapse event. Various ML methods including support vector machines, random forests, and neural networks in combination with fusion methods were used to predict relapse events. After training the above methods on 67% of the participants and testing them on the remaining ones, it was determined that feature fusion techniques applied to actigraphy and EMA metrics produced the highest area under the receiver operating characteristic curve of 0.72. Another novel technique combining raw actigraphy counts with the aforementioned metrics yielded areas of up to 0.65. Overall, the results show that there is much promise that can be held in the integration of digital phenotypes for the prediction of relapse events in adolescents.

In the digital age, the communication gap between parents and adolescents has increased, presenting challenges to understanding the emotional well-being of their children. With the increasing prevalence of social networks, adolescents tend to express their feelings and struggles online rather than engage in face-to-face interaction. Existing monitoring tools allow parents to read messages and observe social media activity, but often fail to interpret the emotional content. Recent studies have explored the feasibility of using natural language processing and machine learning to predict depression based on social media activity. By analyzing the linguistic patterns, sentiments, and emotional content of online communication, researchers have demonstrated the potential to identify individuals suffering from depression at an early stage. This article proposes a novel solution that uses large language models (LLMs) to monitor and analyze adolescent communication on digital platforms, including smartphones and social media. The system aims to detect emotional distress, signs of depression, and other mental health indicators, providing timely alerts to parents. This technology enables parents to understand their teens’ emotions, offer the necessary support, and prevent the escalation of anxiety and depression.

Depression is a common mental disease that has a tendency to develop at a younger age. Early detection of depression with psychological intervention may effectively prevent youth suicide. The establishment of the computer-aided model may be efficient for early detection. However, the existing methods of automatic detection for depression mostly rely on unimodal data. Clinical research shows that patients with depression have specificity in speech, text, expression, and other modal data. Multimodal machine learning is emerging but not yet widely used for the detection of psychiatric disorders. The problem of existing multimodal detection models is that only global or local information is considered in feature fusion, which leads to the low accuracy of the depression detection model. Therefore, this study constructs an automatic detection model based on multimodal machine learning for adolescent depression. The proposed method first extracted four features from audio and text globally and locally; then construct a coarse-grained fusion model and fine-grained fusion model base on these four features; and fuse the coarse-grained and the fine-grained fusion model finally. Experiments on the real-world dataset demonstrate that the proposed method could improve the accuracy of depression detection automatically.

Objective Conventional scale-based diagnostic approaches are increasingly insufficient for addressing the growing mental health challenges among adolescents. Leveraging advances in artificial intelligence, this study aims to develop an accurate, efficient, and scalable model for early identification of adolescent depression risk using large-scale census data, and to identify key daily life factors associated with mental health outcomes. Methods Data were obtained from the 2021 National Survey of Children's Health, including 50,892 adolescents and 463 variables. Based on prior literature, 60 relevant variables were selected. Three progressively structured hypotheses concerning the relationships between adolescent depression and developmental environments were proposed. Machine learning models, including decision trees, XGBoost, support vector machines, and neural networks, were applied to predict depression risk. Mediation analysis was conducted to examine the pathways through which living conditions influence mental health. Results The optimal model demonstrated strong predictive performance, achieving an accuracy of 0.85 and an AUC exceeding 0.87. Feature importance analysis identified several key predictors. Mediation analysis indicated that living conditions exerted a direct effect of 0.225 on mental health, while physical activity and diet quality partially mediated this relationship. Conclusion Living conditions are critical indicators for early identification of adolescent depression risk. The use of nationwide census data enables timely screening and targeted intervention. Improving dietary habits and increasing physical activity may serve as effective preventive strategies for adolescent mental health disorders.

Treating adolescent depression effectively requires providing interventions that are optimally suited to patients’ individual characteristics and needs. Therefore, we aim to develop an algorithm that matches patients with optimal treatment among cognitive-behavioral therapy (CBT), fluoxetine (FLX), and combination treatment (COMB). We leveraged data from a completed clinical trial, the Treatment for adolescents with depression study, where a wide range of demographic, clinical, and psychosocial measures were collected from adolescents diagnosed with major depressive disorder prior to treatment. Machine-learning techniques were employed to derive a model that predicts treatment response (week 12 children’s depression rating scale-revised [CDRS-R]) to CBT, FLX, and COMB. The resulting model successfully identified subgroups of patients that respond preferentially to specific types of treatment. Specifically, our model identified a subgroup of patients (25%) that achieved on average a 16.9 point benefit on the CDRS-R from FLX compared to CBT. The model also identified a subgroup of patients (50%) that achieved an average benefit up to 19.0 points from COMB compared to CBT. Physical illness and disability were identified as overall predictors of response to treatment, regardless of treatment type, whereas baseline CDRS-R, psychosomatic symptoms, school missed, view of self, treatment expectations, and attention problems determined the patients’ response to specific treatments. The model developed in this study provides a critical starting point for personalized treatment planning for adolescent depression.

No abstract available

Human Depression Prediction is essential for several reasons, primarily centered around improving mental health outcomes and providing timely interventions. Firstly, early detection of depression allows for prompt and targeted interventions, enabling individuals to receive appropriate support and treatment before their condition worsens. The Human Depression Prediction Scheme (HDPS) introduces an innovative method for predicting depression by synergizing the capabilities of deep learning through the LeNet architecture with the optimization prowess of the Grey Wolf Optimization (GWO) algorithm. The proposed scheme achieves a commendable accuracy of 95%, attesting to its effectiveness in discerning depressive tendencies. Notably, the HDPS is implemented in Google Colab, emphasizing its accessibility and ease of use. This integration of advanced technologies and optimization techniques positions HDPS as a promising tool for early detection of depression, offering both high accuracy and practical implementation in a widely accessible computing environment.

Depression has become the prevailing global mental health concern. The accuracy of traditional depression diagnosis methods faces challenges due to diverse factors, making primary identification a complex task. Thus, the imperative lies in developing a method that fulfills objectivity and effectiveness criteria for depression identification. Current research underscores notable disparities in brain activity between individuals with depression and those without. The Electroencephalogram (EEG), as a biologically reflective and easily accessible signal, is widely used to diagnose depression. This article introduces an innovative depression prediction strategy that merges time-frequency complexity and electrode spatial topology to aid in depression diagnosis. Initially, time-frequency complexity and temporal features of the EEG signal are extracted to generate node features for a graph convolutional network. Subsequently, leveraging channel correlation, the brain network adjacency matrix is employed and calculated. The final depression classification is achieved by training and validating a graph convolutional network with graph node features and a brain network adjacency matrix based on channel correlation. The proposed strategy has been validated using two publicly available EEG datasets, MODMA and PRED+CT, achieving notable accuracy rates of 98.30 and 96.51%, respectively. These outcomes affirm the reliability and utility of our proposed strategy in predicting depression using EEG signals. Additionally, the findings substantiate the effectiveness of EEG time-frequency complexity characteristics as valuable biomarkers for depression prediction.

The automatic assessment of depression through human voice has gained increasing interest due to its cost-effectiveness and non-invasiveness. This paper employs acoustic embeddings from emotionally rich speech segments for depression prediction, given the strong connection between depression and emotion expression. We leverage a large pre-trained model to predict depression. The public dimensional emotion model (PDEM) used in this study is fine-tuned for recognizing arousal, valence and dominance. We use PDEM for both extracting embeddings and selecting emotionally rich speech segments based on its arousal, valence, and dominance predictions. We advance the state-of-the-art performance on both the Androids corpus (Interview task) for depression detection, following the predetermined protocol, and the E-DAIC corpus for acoustic-based depression severity prediction, adhering to the 2019 Audio Visual Emotion Challenge (AVEC) protocol. The analysis demonstrates that emotionally rich speech segments contain more depression-related cues compared to emotion-neutral segments.

Trustworthy depression prediction based on deep learning, incorporating both predictive reliability and algorithmic fairness across diverse demographic groups, is crucial for clinical application. Recently, achieving reliable depression predictions through uncertainty quantification has attracted increasing attention. However, few studies have focused on the fairness of uncertainty quantification (UQ) in depression prediction. In this work, we investigate the algorithmic fairness of UQ, namely Equal Opportunity Coverage (EOC) fairness, and propose Fair Uncertainty Quantification (FUQ) for depression prediction. FUQ pursues reliable and fair depression predictions through group-based analysis. Specifically, we first group all the participants by different sensitive attributes and leverage conformal prediction to quantify uncertainty within each demographic group, which provides a theoretically guaranteed and valid way to quantify uncertainty for depression prediction and facilitates the investigation of fairness across different demographic groups. Furthermore, we propose a fairness-aware optimization strategy that formulates fairness as a constrained optimization problem under EOC constraints. This enables the model to preserve predictive reliability while adapting to the heterogeneous uncertainty levels across demographic groups, thereby achieving optimal fairness. Through extensive evaluations on several visual and audio depression datasets, our approach demonstrates its effectiveness.

Major depressive disorder (MDD) remains a leading cause of disability worldwide, yet current diagnostic approaches rely heavily on subjective clinical assessments, which can delay intervention and reduce diagnostic reliability. Objective, multimodal biomarkers offer a promising route toward earlier and more accurate detection. Electroencephalography (EEG) captures real-time neural dynamics, while genetic biomarkers provide stable indicators of molecular predisposition, offering complementary perspectives on depression risk. This study introduces EEG–Gene Fusion Depression Network (EGF-DepNet), the first end-to-end deep learning framework to integrate EEG-derived features and gene-based biomarkers for depression prediction. Two fusion strategies are proposed: (i) an Attention Fusion model employing cross-modal MultiHeadAttention within a compact Conv1D network, and (ii) a Transformer Fusion model that encodes EEG and genomic embeddings as tokens in a lightweight self-attention encoder. Using the ICBrainDB dataset, both architectures achieved high predictive performance, with Transformer Fusion outperforming Attention Fusion across multiple evaluation metrics, including F1-score (0.727 vs. 0.600) and AUC (0.845 vs. 0.749). Results demonstrate that multimodal EEG–genomic integration improves classification robustness over unimodal approaches, effectively leveraging the temporal sensitivity of EEG and the trait stability of genetic markers. This work advances the development of biologically informed, AI-driven diagnostic tools, offering a pathway toward more precise, scalable, and personalized approaches in precision psychiatry.

This paper focuses on the adaptive optimization problem of transfer learning in cross-sample electroencephalogram depression prediction models. This paper expounds the significance of electroencephalogram (EEG) signals in depression prediction and the challenges faced in cross-sample prediction, and analyzes the basic concepts and common methods of transfer learning and its preliminary application in the medical field. This paper explores the influence of factors such as data differences and individual differences on the model in cross-sample electroencephalogram (EEG) depression prediction, and elaborates in detail the adaptive optimization strategies of transfer learning in data preprocessing, feature extraction, model training and adjustment, etc. It points out the current challenges and looks forward to the future development direction, aiming to provide theoretical support for constructing a more accurate and universal cross-sample electroencephalogram depression prediction model.

No abstract available

The severity of depression among young Australians cannot be overstated, as it continues to have a profound impact on their mental health and general wellbeing. This study used machine learning (ML) algorithms to analyse longitudinal data, identifying key features to predict depression, assess future risk, and explore age-specific behaviours that contribute to its progression over time. The results emphasize the significance of early detection to prevent unfavourable consequences and shed light on the alterations in depressive symptoms during various stages of development. Three widely regarded ML techniques—random forest (RF), support vector machine (SVM), and logistic regression (LR)—are being applied and compared with a longitudinal data analysis. Additionally, the Apriori algorithm is being utilized to explore potential relationships between health, behaviour, and activity issues with depression among different age groups (10–17). The analysis results indicate that the RF model is performing exceptionally well in diagnosing depression, with a 94% accuracy rate and weighted precision of 95% for non-depressed and 88% for depressed cases. In addition, the LR model shows promising results, achieving an 89% accuracy rate and 91% weighted precision. Moreover, insights from the Apriori algorithm underscore the significance of early detection by examining potential associations between health, behaviour, and activity problems and depression across diverse age groups. Combining early screening programs with the RF model and the Apriori algorithm is crucial for understanding depression and developing effective prevention strategies. Emphasizing Apriori's factors and regularly updating strategies with new information will enhance depression management and prevention.

The topics covered in this article include the creation of a bootstrap learning model for depression predictions based on the Random Forest technique. Depression is a severe mental illness that affects millions of people worldwide. This condition causes great pain and affects quality of life, and in the most severe cases, the person takes their own life. Depression has a high incidence, but diagnosis is always complex and often delayed. It is made on the basis of clinical assessment, which is subjective, and questionnaires, which are often inaccurate and cannot identify people at risk early enough because a person's subjective perception can often be distorted. In this context, and to illustrate our point, we aim to show how AI, and more specifically machine learning, can provide innovative applications that can be used to improve early detection cf. prevention of depression risk. Instead of stupidly defining score intervals for a child, we can train a model on a dataset to identify patterns and correlations that escape simple regression analyses. Then, we can anticipate the first signs of log-in with depression, or we can identify which combinations of self and family history are most concerning. To complement our study, we chose the decision tree ensemble algorithm. The article highlights the need for more objective and effective prediction tools for depression, and proposes a machine learning-based solution to achieve this, potentially leading to earlier intervention and better patient care.

Automatic detection of depression is a rapidly growing field of research at the intersection of psychology and machine learning. However, with its exponential interest comes a growing concern for data privacy and scarcity due to the sensitivity of such a topic. In this paper, we propose a pipeline for Large Language Models (LLMs) to generate synthetic data to improve the performance of depression prediction models. Starting from unstructured, naturalistic text data from recorded transcripts of clinical interviews, we utilize an open-source LLM to generate synthetic data through chain-of-thought prompting. This pipeline involves two key steps: the first step is the generation of the synopsis and sentiment analysis based on the original transcript and depression score, while the second is the generation of the synthetic synopsis/sentiment analysis based on the summaries generated in the first step and a new depression score. Not only was the synthetic data satisfactory in terms of fidelity and privacy-preserving metrics, it also balanced the distribution of severity in the training dataset, thereby significantly enhancing the model's capability in predicting the intensity of the patient's depression. By leveraging LLMs to generate synthetic data that can be augmented to limited and imbalanced real-world datasets, we demonstrate a novel approach to addressing data scarcity and privacy concerns commonly faced in automatic depression detection, all while maintaining the statistical integrity of the original dataset. This approach offers a robust framework for future mental health research and applications.

While existing depression prediction methods based on deep learning show promise, their practical application is hindered by the lack of trustworthiness, as these deep models are often deployed as black box models, leaving us uncertain on the confidence of their predictions. For high-risk clinical applications like depression prediction, uncertainty quantification is essential in decision-making. In this paper, we introduce conformal depression prediction (CDP), a depression prediction method with uncertainty quantification based on conformal prediction (CP), giving valid confidence intervals with theoretical coverage guarantees for the model predictions. CDP is a plug-and-play module that requires neither model retraining nor an assumption about the depression data distribution. As CDP provides only an average coverage guarantee across all inputs rather than per-input performance guarantee, we further propose CDP-ACC, an improved conformal prediction with approximate conditional coverage. CDP-ACC firstly estimates the prediction distribution through neighborhood relaxation, and then introduces a conformal score function by constructing nested sequences, so as to provide a tighter prediction interval adaptive to specific input. We empirically demonstrate the application of CDP in uncertainty-aware facial depression prediction, as well as the effectiveness and superiority of CDP-ACC on the AVEC 2013 and AVEC 2014 datasets.

Human Depression Prediction is important for a number of reasons, the most important of which are that it can help with faster interventions and better mental health outcomes. First and foremost, early identification of depression facilitates timely and focused interventions, allowing patients to obtain the right care and support prior to the worsening of their illness. By combining the deep learning powers of the LeNet architecture with the optimization skills of the Grey Wolf Optimization (GWO) algorithm, the Human Depression Prediction Scheme (HDPS) presents a novel approach to depression prediction. The proposed technique shows promise in identifying depression tendencies with a noteworthy 95% accuracy rate.Notably, Google Colab uses the HDPS, highlighting its usability and accessibility. By combining cutting-edge technologies and optimization strategies, HDPS is positioned as a potentially useful tool for depression early diagnosis, providing high accuracy and useful application in a computing environment that is widely available.

Abstract. With the acceleration of modern society's pace, depression has become a common mental health issue. This study aims to develop a depression detection model based on natural language processing (NLP) technology to automatically identify potential depression patients. First, collect data from the Weibo platform using web scraping techniques, and then use NLP tools such as XLNet for in-depth analysis of the text data. In the model evaluation phase, the paper uses key metrics such as accuracy (ACC) to assess the effectiveness of NLP techniques in predicting depression. Experimental results indicate that the XLNet model performed the best among all tested models, achieving an accuracy of 81%, thus confirming the feasibility of using NLP technology for depression prediction. The significance of this study lies in providing a new method for mental health monitoring, which aids in the early detection and intervention of depression, with important social value and application prospects.

In recent years, the increasing number of individuals diagnosed with depression and the growing awareness of its impact on modern society have highlighted the significance of accurate depression diagnosis. Microarray data has played a crucial role in uncovering the genetic mechanisms underlying depression. However, existing methods for depression prediction using microarray data often rely on the selection of differentially expressed genes. This approach disregards important information from other genes and is susceptible to batch effects, thereby limiting generalizability and model stability. To address these limitations, we propose DP-BERT, a depression prediction model based on Bidirectional Encoder Representations from Transformers (BERT). DP-BERT follows a pre-training and fine-tuning paradigm, leveraging a large amount of unlabeled microarray data from diverse sequencing platforms for pretraining to extract comprehensive genetic-level representations of psychiatric disorders. Subsequently, supervised fine-tuning is performed for depression prediction. Experimental results demonstrate that the pre-trained model achieves superior performance in depression prediction. The source code can be obtained from https://github.com/CSUBioGroup/DP-BERT.

Depression is a multifaceted mental illness that has a significant impact on a person's thoughts, feelings, and general state of well-being. This presents a serious problem for students which not only affects their academic achievement but also affects their mental health. A panoptic analysis of machine learning techniques for depression prediction is presented in this paper. Through a Google survey targeting university students, we collected a dataset of 550 examples covering a wide range of relevant features. Five widely used machine learning models were employed after efficient data preprocessing and feature selection: Logistic Regression, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision trees, and Neural networks. After complete evaluation of all pre-trained models among them Neural network outperforms others with the accuracy of 94.55%. Based on this success, an improved combination method including SVM, KNN and Logistic Regression with a Neural network as a meta-model was developed. This stacking ensemble learning method provided an outstanding accuracy of 96.36%, surpassing the performance of individual models. The proposed model pushes the boundaries of machine learning in mental health analytics, demonstrating tremendous potential as a powerful solution for depression prediction.

Depression affects millions worldwide, emphasizing the need for early detection. Leveraging machine learning, our research introduces a novel deep learning model merging text and social media data for depression prediction. Comparative analysis with state-of-the-art methods demonstrates promising results. As heightened social media use correlates with increased depression rates, our study targets probable depressed Twitter users through machine learning. By analyzing both network behavior and tweets, we develop classifiers utilizing diverse features extracted from user activities, revealing that incorporating more features enhances accuracy and F-measure scores in identifying depressed users. Our data-driven approach offers a predictive tool for early depression detection and other mental illnesses. This paper contributes insights into depression detection using machine learning and proposes innovative strategies for improved diagnosis and treatment

According to World Health Organisation (WHO), Depression is a leading cause of disability worldwide with almost percent of the population affected, including 5.0 percent of adults. There are nearly 280 million people in the world who suffer from depression. Major depression disorder (MDD) is the greatest cause of disability and morbidity, post-COVID-19. It is known to significantly affect the quality of life and create a socioeconomic burden. The solution to this problem includes diagnosing depression before even the onset of the disease. There exist many cases studies have been conducted in this area of research which includes clinical trials on using DTI-MRI Images and MRI Images to detect depression in mood-affected and non-affected individuals. It was proven to be successful that depression can be diagnosed from DTI-MRI and MRI Images. This served as an innovation towards this research of automating depression prediction from DTI-MRI Images using Machine Learning and thus saving lives.

Abstract Background Depression represents a pressing global public health concern, impacting the physical and mental well-being of hundreds of millions worldwide. Notwithstanding advances in clinical practice, an alarming number of individuals at risk for depression continue to face significant barriers to timely diagnosis and effective treatment, thereby exacerbating a burgeoning social health crisis. Objective This study seeks to develop a novel online depression risk detection method using natural language processing technology to identify individuals at risk of depression on the Chinese social media platform Sina Weibo. Methods First, we collected approximately 527,333 posts publicly shared over 1 year from 1600 individuals with depression and 1600 individuals without depression on the Sina Weibo platform. We then developed a hierarchical transformer network for learning user-level semantic representations, which consists of 3 primary components: a word-level encoder, a post-level encoder, and a semantic aggregation encoder. The word-level encoder learns semantic embeddings from individual posts, while the post-level encoder explores features in user post sequences. The semantic aggregation encoder aggregates post sequence semantics to generate a user-level semantic representation that can be classified as depressed or nondepressed. Next, a classifier is employed to predict the risk of depression. Finally, we conducted statistical and linguistic analyses of the post content from individuals with and without depression using the Chinese Linguistic Inquiry and Word Count. Results We divided the original data set into training, validation, and test sets. The training set consisted of 1000 individuals with depression and 1000 individuals without depression. Similarly, each validation and test set comprised 600 users, with 300 individuals from both cohorts (depression and nondepression). Our method achieved an accuracy of 84.62%, precision of 84.43%, recall of 84.50%, and F1-score of 84.32% on the test set without employing sampling techniques. However, by applying our proposed retrieval-based sampling strategy, we observed significant improvements in performance: an accuracy of 95.46%, precision of 95.30%, recall of 95.70%, and F1-score of 95.43%. These outstanding results clearly demonstrate the effectiveness and superiority of our proposed depression risk detection model and retrieval-based sampling technique. This breakthrough provides new insights for large-scale depression detection through social media. Through language behavior analysis, we discovered that individuals with depression are more likely to use negation words (the value of “swear” is 0.001253). This may indicate the presence of negative emotions, rejection, doubt, disagreement, or aversion in individuals with depression. Additionally, our analysis revealed that individuals with depression tend to use negative emotional vocabulary in their expressions (“NegEmo”: 0.022306; “Anx”: 0.003829; “Anger”: 0.004327; “Sad”: 0.005740), which may reflect their internal negative emotions and psychological state. This frequent use of negative vocabulary could be a way for individuals with depression to express negative feelings toward life, themselves, or their surrounding environment. Conclusions The research results indicate the feasibility and effectiveness of using deep learning methods to detect the risk of depression. These findings provide insights into the potential for large-scale, automated, and noninvasive prediction of depression among online social media users.

Language provides significant insights into an individual’s emotional state, social status, and personality traits. This research aims to enhance depression detection through the analysis of linguistic features and various dataset attributes. The dataset, sourced from the social networking platform Reddit, comprises posts and comments from individuals diagnosed with depression. Logistic regression with term frequency-inverse document frequency (TF-IDF) is employed as the primary model for text classification. To improve model performance, a novel feature—the average time interval between consecutive posts or comments—is introduced, contributing to a marginal but noteworthy improvement in accuracy. The proposed model demonstrates superior F1 scores compared to other models applied to the same dataset. Given the increasing recognition of mental health’s significance, accurately diagnosing mental disorders is of paramount importance. This study underscores the potential of leveraging linguistic analysis and advanced machine learning techniques to identify depressive symptoms, thereby contributing to more effective mental health diagnostics and interventions.

No abstract available

Depression is a common and severe mental disorder that frequently goes undiagnosed and untreated, particularly during its initial phase. However, with the increasing number of people sharing their thoughts and feelings online, social media has become a valuable resource to identify symptoms of mental health problems such as depression. As a result, research on social media-based depression diagnosis has received significant interest but it mainly focuses on the semantics of posts. It cannot detect the implicit ambiguities that lies in the language of the user in an online posting. To address this limitation, a syntactical analysis of each user post is required. The goal of this research is to enhance the identification of individuals at risk of depression by analyzing their language and rhetorical relations using rhetorical structure theory. The rhetorical scores are incorporated into an ensemble machine learning model to classify social media posts as depressive or non-depressive. The model employs a combining multilayer perceptron, extreme gradient boosting and support vector machine algorithms. Therefore we proposed a model named RSTFusionX by linking depression to a social media dataset and using a combination of the rhetorical structure theory framework along with an ensemble machine learning algorithm. To attain more precise prediction models, the research aims to better understand the association between social media language trends and depression. The results of this study are significant, as RSTFusionX achieves 97.1% accuracy, 97.4% precision, 96.7% recall and 96.9% F1-score, exhibiting better results as compared to baseline algorithms. As a result of RSTFusionX, true positive and true negative values are increased while false positive and false negative predictions are reduced. The proposed research findings suggest that the integration of characteristics from the depressive vocabulary improves classification accuracy. The article contributes to the resources for depression detection and provides a technique for future research to further improve depression detection.

No abstract available

Depression is a mental disorder characterized by a sad mood, irritability, anger, agitations, loss of interest or pleasure, reduced energy, feelings of guilt, low self-esteem, troubled sleep, appetite loss, and poor attentiveness. The effects of late diagnosis of depression in Nigerian students have posed threats to the academic performance of the students, economic growth, and security threats. To address this challenge, an ANFIS model for early detection of depression among Nigerian Students is proposed. This aids in the reduction and possible elimination of prevalent cases of depression-related dangers among students in tertiary institutions. ANFIS is utilized because of its transparency and ability to classify and identify hidden symptoms of depression, and its tendency for reduced memorization errors for users. The database was developed to hold user data, symptoms, and prescriptions and linked to the ANFIS framework to enable the diagnosis of early-phase depression. Data was collected from the University of Uyo primary health care center, and the University of Uyo Teaching Hospital (UUTH). The ANFIS model implementation was implemented in MATLAB while the application forming the input interface was implemented with JAVA. The dataset for training was passed through ANFIS for 10 epochs and upon completion the system had a training error of 6.0138e-0.5 and an average testing error of 4.6648 on the test data, these results indicate that the system possessed 95% classification accuracy in the detection of early depression in Nigerian students.

No abstract available

Abstract Objective: The purpose of this study was to screen pertinent variables to identify ordered relations that provide easily interpretable and accurate predictions of the probability of cigarette use among adolescents using a classification tree modeling approach. Methods: This cross-sectional study included a national sample of 3717 U.S. adolescents aged between 12 and 20 years old from the 2016 National Survey on Drug Use and Health. Results: The results indicated that age was the most influential variable, followed by depression, race/ethnicity, family income, gender, and alcohol abuse and dependence. Additionally, several interaction emerged that identified higher and lower cigarette use profiles: youth who were between 18 and 20 years old and self-identified as non-Hispanic White, Native American/Alaska Native, and “Other” racial/ethnic group and African American, Asian, and Latinx adolescents, with depressive symptoms were at higher risk of cigarette use; while youth who reported lower family incomes, were 16–17 years old, who identified as African American, Asian, and Latinx, were also likely to report lower use of cigarettes when they reported lower depressive symptom scores. Discussion: These results are discussed relative to practice implications.

Depressive disorder, frequently referred to as depression, is a common mental health condition that involves persistent feelings of sadness and loss of interest in daily activities. The use of objective techniques for the diagnosis and monitoring of this condition can significantly facilitate its management, in complement to traditional self-reports and clinical assessments. Based on these considerations, this study presents a preliminary investigation into the potential of electrodermal activity (EDA) in recognizing depressive symptoms. Additionally, to standardize the experimental protocol, emotional Virtual Humans are used to interact with subjects through conversations, thereby facilitating a controlled and repeatable environment. The aim is to develop an emotional recognition model to assess depressive symptoms, supporting clinicians and sufferers in diagnosing and monitoring depression. The experimental findings show promising results, suggesting that EDA holds the potential to facilitate a deeper understanding of depressive symptoms. This can lead to improved diagnostic accuracy and personalized treatment approaches for depression management.

A significant obstacle to women's lives is menstruation, which, in some cases, leads to depression, anxiety, and stress. This research, therefore, aims to solve this problem with three objectives: to study the context of female adolescents who experience premenstrual syndrome (PMS) problems and depression crises, to develop a classification model for personalized premenstrual symptoms and depressive crisis risk tracking using data analytics and machine learning techniques, and to evaluate the performance of all developed models to select suitable models for future application. The research data included 282 volunteers who gave consent from the University of Phayao, with random sampling. The machine learning used in this research consists of six techniques: decision trees, k-nearest neighbors, logistic regression, naïve bayes, random forests, and support vector machines. The model performance is tested using 10-fold cross-validation and evaluated using five metrics: accuracy, precision, recall, f1-score, and time. The results of the construction and evaluation of all models are at a high level, which is encouraging. The researchers expect that this research can be further developed for future mobile applications.

Monitoring human behavior through wearable devices has potential in psychiatry. Among them, actigraphy data has been used to classify depression and detect depressive symptoms. We aim to collect a larger number of data to measure classification performance. This study evaluates the performance of classifying depressive symptoms solely on actigraphy data using both public (n=1549) and collected (n=3145) datasets. We found that there are challenges in classifying depressive symptoms from actigraphy data.

Major Depressive Disorder (MDD) is a complicated mental illness that consists of wide range of correlated symptoms. According to DSM-V, it is characterized by pervasive low mood, loss of interest or pleasure in nearly all activities, and additional symptoms that can cause significant distress in social, occupational, and important areas of life. In this study, association rules, decision tree classification and agglomerative clustering are employed to classify MDD symptoms interconnection and their co-occurrence pattern. A combination of key symptoms occurrence, such as aggression with euphoric responses and overthinking with euphoria are identified through association rules that show highest lift values. Decision tree is employed to predict primary node which is mood swing as key predictor of MDD. Then, agglomerative clustering is used to split the dataset into three clusters based on expert diagnosis to identify the range of symptoms that overlap. In this study, computational approach is utilized to unravel the hidden pattern and overlapping of correlated symptoms that will help in improved diagnosis and better personalized treatment plans. This study highlighted the interrelationships of symptoms associated with MDD and their thorough examination for therapeutic approach. Future perspectives should focus on diverse datasets with extension and validation of findings that produce sustainable findings for clinical decision making. By the complex interplay of symptoms of MDD shows the contribution of this research towards advancement of evidence-based diagnostics, ultimately aim to improve clinical outcomes by intergradational symptomatic study.

The diagnosis and classification of depressive disorders pose significant challenges in mental healthcare, mainly due to overlapping symptoms, subjective evaluations, and variations in patient presentations. Traditional diagnostic approaches often lack objectivity and fail to capture the complex nature of depression across diverse populations. This study introduces a comprehensive framework that leverages advanced Machine Learning (ML) and Deep Learning (DL) models to improve the accuracy and reliability of diagnosing depressive disorders. Using the SAMM (Spontaneous Micro-Facial Movement) dataset, comprising 11,800 high-resolution facial images capturing spontaneous facial expressions, the proposed framework integrates dual embedding methods (GloVE and BERT) with hierarchical attention mechanisms for feature extraction. Parallel processing streams of LSTM and CNN architectures allow the recognition of intricate patterns across multimodal data. Experimental results showed superior performance across key metrics, achieving an accuracy of 94%, precision of 92%, recall of 93%, F1-score of 92.5%, and an AUC-ROC of 0.96. The proposed framework provides an efficient, interpretable, and scalable solution to advance mental health diagnostics, addressing the urgent need for objective and standardized tools in psychiatric care.

This study focuses on classifying depression from Twitter posts using deep learning techniques, particularly through the Deep learning model. Depression is a critical issue impacting both individuals and society, with potentially severe consequences, including mortality. The study leverages social media data to identify signs of depression, offering a technological approach to understanding and predicting depressive behaviors. By collecting English text data, including hash-tagged tweets indicative of depressive symptoms, and classifying data into nine categories of depressive symptoms, the study enhances the accuracy and prediction capacity of the model. The proposed classification system aims to support efforts in identifying and providing early intervention for individuals at risk of depression.

Early detection of mental depression prevents the severe repercussions of long-term depressive symptoms such as suicidal thoughts and ideation. With the widespread use of social media and the internet these days, prompt identification of emotional reactions is essential. Therefore, monitoring social media texts like Facebook comments or tweets could be highly helpful in detecting the mental depression.

No abstract available

Depression is a common illness worldwide with potentially severe implications. Early identification of depressive symptoms is a crucial first step towards assessment, intervention, and relapse prevention. With an increase in data sets with relevance for depression, and the advancement of machine learning, there is a potential to develop intelligent systems to detect symptoms of depression in written material. This work proposes an efficient approach using Long Short-Term Memory (LSTM)-based Recurrent Neural Network (RNN) to identify texts describing self-perceived symptoms of depression. The approach is applied on a large dataset from a public online information channel for young people in Norway. The dataset consists of youth’s own text-based questions on this information channel. Features are then provided from a one-hot process on robust features extracted from the reflection of possible symptoms of depression pre-defined by medical and psychological experts. The features are better than conventional approaches, which are mostly based on the word frequencies (i.e., some topmost frequent words are chosen as features from the whole text dataset and applied to model the underlying events in any text message) rather than symptoms. Then, a deep learning approach is applied (i.e., RNN) to train the time-sequential features discriminating texts describing depression symptoms from posts with no such descriptions (non-depression posts). Finally, the trained RNN is used to automatically predict depression posts. The system is compared against conventional approaches where it achieved superior performance than others. The linear discriminant space clearly reveals the robustness of the features by generating better clustering than other traditional features. Besides, since the features are based on the possible symptoms of depression, the system may generate meaningful explanations of the decision from machine learning models using an explainable Artificial Intelligence (XAI) algorithm called Local Interpretable Model-Agnostic Explanations (LIME). The proposed depression symptom feature-based approach shows superior performance compared to the traditional general word frequency-based approaches where frequency of the features gets more importance than the specific symptoms of depression. Although the proposed approach is applied on a Norwegian dataset, a similar robust approach can be applied on other depression datasets developed in other languages with proper annotations and symptom-based feature extraction. Thus, the depression prediction approach can be adopted to contribute to develop better mental health care technologies such as intelligent chatbots.

No abstract available

Depressive episode is key symptom collection of mood disorders. Early intervention can prevent it from happening or reduce its impact, and close monitoring can greatly improve medical management. However, most current monitoring methods are ex post facto, coarse in time granularity and resource consuming. In this article, we aimed to develop a cost-friendly and high usability depressive episode detection framework. In Phase I, we fitted instantaneous affective state models by using R-R intervals collected with photoplethysmogram sensors in smartwatches from laboratory experiments of 1107 participants. In Phase II we utilized the models from Phase I to record long-term affective experience of 2192 participants. Depressive episode models were fitted with affective experience time series. The best instantaneous affective states models achieved overall accuracies of 91% with 2 classes (neutral/ aroused) and 82% with 3 classes (joy/ neutral/ sadness), and the depressive episode models (less severe/ more severe) achieved an overall accuracy of 76% and a best accuracy of 88%. We investigated and discussed the performance differences of the models with multiple settings. We found person-based feature normalization is effective in improving model performance for subjective affect experience. We also found identification of diurnal mood variation may be critical in depressive episode detection.

Studies on intergenerational relationships between parents and children in Asian American families highlight their impact on mental health and well-being. This study investigates the role of ambivalent emotions in online narratives shared by Asian and Asian American children on the subreddit, r/Asianparentstories. By employing a BERT-based model to detect emotion at the sentence level and depressive symptoms at the post level, we analyze mixed feelings to better understand how they predict depressive symptoms. First, among 28 detectable, eight (realization, approval, sadness, anger, curiosity, annoyance, disappointment, disapproval) comprise over 50%, exhibiting significant co-occurrence among themselves and with other emotions. Second, we find the co-occurrence of multiple emotions, indicating that emotions in a single post are not limited to consistently positive or negative feelings. Finally, our findings indicate that while negative emotion pairs (e.g., confusion-grief, anger-grief) are associated with depressive symptoms, positive emotion pairs (e.g., admiration-realization, amusement-joy) negatively correlate with depressive symptoms, and combinations of ambivalent emotions indicate varied results in predicting depressive symptoms. These findings highlight the importance of automated emotion classification and the need to consider emotional ambivalence, which holds practical and clinical implications for understanding the dynamics of parent-child relationships.

Patients with severe depressive symptoms are exposed to high suicidal risk. Since assessing depressive symptoms is either implicated by variability among subjects and clinicians or by time-consuming procedures, it is difficult to do objectively and effortlessly. There is a need for automatic approaches to estimate them. Previous investigations on speech-based severity’s categorical assignment predominantly focused on distinguishing depressive disorders from healthy. This article presents an automatic assessment system of severe self-reported depressive symptoms (SSDSs) to classify speakers with severe depressive symptoms from normal and not so severe ones. First, modulation-domain spectral centroid mean (MSCM) features characterizing severe depressive symptoms are extracted. Subsequently, aiming at the problems of limited and imbalanced training data, we establish variance-weighted sum-to- ${H}$ constraint ( ${H}$ is an adjustable parameter) collaborative representation-based classification (VWSC-CRC) method to exploit interclass separability between SSDS and non-SSDS. The proposed system (MSCM–VWSC-CRC) was evaluated on the small and imbalanced data set from AVEC2013 by dominant metrics, such as F1-score and area under the precision–recall curve (AU-PRC), as well as auxiliary indices, including specificity, recall (sensitivity), precision, accuracy, and area under the receiver operating characteristic curve (AU-ROC) as needed. The results exhibit a clear advantage over three important systems in the literature. The gains of this article are likely to lay the substantive groundwork to assist clinicians in automatically screening subjects with severe depressive symptoms so as to facilitate the diagnosis of the corresponding psychological disorders.

Background Previous studies have classified major depression and healthy control groups based on vocal acoustic features, but the classification accuracy needs to be improved. Therefore, this study utilized deep learning methods to construct classification and prediction models for major depression and healthy control groups. Methods 120 participants aged 16–25 participated in this study, included 64 MDD group and 56 HC group. We used the Covarep open-source algorithm to extract a total of 1200 high-level statistical functions for each sample. In addition, we used Python for correlation analysis, and neural network to establish the model to distinguish whether participants experienced depression, predict the total depression score, and evaluate the effectiveness of the classification and prediction model. Results The classification modelling of the major depression and the healthy control groups by relevant and significant vocal acoustic features was 0.90, and the Receiver Operating Characteristic (ROC) curves analysis results showed that the classification accuracy was 84.16%, the sensitivity was 95.38%, and the specificity was 70.9%. The depression prediction model of speech characteristics showed that the predicted score was closely related to the total score of 17 items of the Hamilton Depression Scale(HAMD-17) (r=0.687, P<0.01); and the Mean Absolute Error(MAE) between the model’s predicted score and total HAMD-17 score was 4.51. Limitation This study’s results may have been influenced by anxiety comorbidities. Conclusion The vocal acoustic features can not only effectively classify the major depression and the healthy control groups, but also accurately predict the severity of depressive symptoms.

Major depressive disorder (MDD) and bipolar disorder (BD) are two major mood disorders with partly overlapped symptoms but different treatments. However, their misdiagnosis and mistreatment are common based on the DSM-V criteria, lacking objective and quantitative indicators. This study aimed to develop a novel approach that accurately classifies MDD and BD based on their resting-state magnetoencephalography (MEG) signals during euthymic phases. A revisited 3D CNN model, Semi-CNN, that could automatically detect brainwave patterns in spatial, temporal, and frequency domains was implemented to classify wavelet-transformed MEG signals of normal controls and MDD and BD patients. The model achieved a test accuracy of 96.05% and an average of 95.71% accuracy for 5-fold cross-validation. Furthermore, saliency maps of the model were estimated using Grad-CAM++ to visualize the proposed classification model and highlight disease-specific brain regions and frequencies. Clinical Relevance - Our model provides a stable pipeline that accurately classifies MDD, BD, and healthy individuals based on resting-state MEG signals during the euthymic phases, opening the potential for quantitative and accurate brain-based diagnosis for the highly misdiagnosed MDD/BD patients.

Mental disorders are widely recognized as a major contributor to the global burden of disease. Heart rate variability (HRV) serves as an objective quantitative measure of autonomic nervous system (ANS) dysregulation and can be used to investigate psychiatric symptoms, including depression and anxiety. The aim of this study is to differentiate the severity of depression and anxiety, two prevalent symptoms of mental disorders, based on HRV. The fundamental deep learning architecture, specifically the Multilayer Perceptron (MLP) network, was employed for the classification of severity of both symptoms. By leveraging deep learning network, the classification of depressive symptoms achieved an accuracy of 83.8%, while the classification of anxious symptoms achieved an accuracy of 78.4%, demonstrating superior discrimination power compared to the conventional machine learning models.

Abstract Background Although voice has been proposed as a potential biomarker for depression detection, standardized biomarkers that can be widely applied in clinical practice remain insufficient. The advent of big data and artificial intelligence analysis have given rise to a significant increase in research on speech depression recognition (SDR), which is aimed at identifying depressive symptoms from voice. However, existing studies are limited by small sample sizes, which constrain the exploration of diverse analytical methods. Aims & Objectives This study aims to investigate how SDR model performance can be improved by constructing balanced datasets that reflect the characteristics of the data, with a focus on data interpretation rather than model fine-tuning. Method A total of 3425 participants were recruited and their voices were recorded as they read a predefined paragraph aloud. We evaluated classification performance by adjusting the thresholds between normal and depressive symptom groups based on PHQ-9 scores, exploring various combinations of these thresholds. Classification was performed using a Random Forest model, complemented by additional models such as Deep Neural Networks and XGBoost, all evaluated with nested cross-validation. Using these imbalanced datasets and a balanced dataset matched to the sample size of the depressive group, we measured classification performance using random forests for each set. Results The classification performance yielded suboptimal results on imbalanced datasets. However, when evaluated on balanced data, the model demonstrated a clearer potential for distinguishing between the groups, suggesting that classification is more feasible under balanced conditions. Discussion & Conclusions To enhance model learning in SDR research, the crucial point may be not only a large sample size but also balancing the dataset based on the severity of depressive symptoms.

Major Depressive Disorder (MDD) is a very serious mental illness that can affect the daily lives of patients. Accurate diagnosis of this disorder is necessary for planning individualized treatment. However, diagnosing MDD requires the clinicians to personally interview the subjects and rate the symptoms based on Diagnostic and Statistical Manual of Mental Disorders (DSM), which can be very time consuming. Discovering quantifiable signals and biomarkers associated with MDD using functional magnetic resonance imaging (fMRI) scans of patients have the potential to assist the clinicians in their assessment. This paper explores the use of resting-state functional connectivity and network features to classify MDD vs. healthy subjects. For each subject, mean time-series are extracted from 85 brain regions and they are decomposed to 4-frequency bands. Mean time-series for each of the frequency bands are utilized to compute the Pearson correlation and network characteristics. Features are selected separately from correlation and network characteristics using Minimum Redundancy Maximum Relevance (mRMR) to create the final classifier. The proposed scheme achieves 79% accuracy (65 out of 82 subjects classified correctly) with 86% sensitivity (42 out of 49 MDD subjects identified correctly) and 70% specificity (23 out of 33 controls identified correctly) using leave-one-out classification with in-fold feature selection. Pearson correlation had the highest discrimination in band 0.015-0.03 Hz and network based features had the highest discrimination in band 0.03-0.06 Hz for distinguishing MDD vs. healthy subjects.

Acoustic features are crucial behavioral indicators for depression detection. However, prior speech-based depression detection methods often overlook the variability of emotional patterns across samples, leading to interference from speaker identity and hindering the effective extraction of emotional changes. To address this limitation, we developed the Emotional Word Reading Experiment (EWRE) and introduced a method combining self-supervised and supervised learning for depression detection from speech called MFE-Former. First, we generate fine-grained emotional representations for response segments by computing cosine similarity between intra-sample and inter-sample contexts. Concurrently, orthogonality constraints decouple identity information from emotional features, while a Transformer decoder reconstructs spectral structures to improve sensitivity to depression-related emotional patterns. Next, we propose a multi-scale emotion change perception module and a Bernoulli distribution-based joint decision module integrate multi-level information for depression detection. By enhancing the distribution differences among positive, neutral, and negative emotional features, we find that patients with depression are more inclined to express negative emotions, whereas healthy individuals express more positive emotions. The experimental results on EWRE and AVEC 2014 show that MFE-Former outperforms state-of-the-art temporal methods under conditions of variability in emotional patterns across samples.

Depression detection from speech faces two critical challenges: the subtle time-frequency feature variations in speech are difficult to comprehensively capture, and the acquisition of annotated data is costly and limited in scale. This paper proposes DepMAE, a self-supervised learning framework based on masked autoencoders and spectrograms for automatic depression detection. The framework preserves complete time-frequency features of speech signals through spectrogram analysis and leverages the self-supervised learning capability of masked autoencoders (with 65% masking rate) to learn depression-related latent feature representations from unlabeled data. We adopt a “pre-training-fine-tuning” paradigm, designing a specialized BDI-II score predictor based on the ViT encoder-decoder architecture. Experiments on the AVEC2014 dataset demonstrate that, compared to existing methods, DepMAE achieves approximately 12% and 8% performance improvements in MAE and RMSE metrics, respectively, reaching state-of-the-art performance. Systematic ablation studies validate the effectiveness of key parameter settings, while visualization analysis confirms that the model successfully captures speech features related to depression, providing a promising approach for automatic depression assessment from speech signals.

A fundamental component of user-level social media language based clinical depression modelling is depression symptoms detection (DSD). Unfortunately, there does not exist any DSD dataset that reflects both the clinical insights and the distribution of depression symptoms from the samples of self-disclosed depressed population. In our work, we describe a semi-supervised learning (SSL) framework which uses an initial supervised learning model that leverages (1) a state-of-the-art large mental health forum text pre-trained language model further fine-tuned on a clinician annotated DSD dataset, (2) a Zero-Shot learning model for DSD, and couples them together to harvest depression symptoms related samples from our large self-curated depressive tweets repository (DTR). Our clinician annotated dataset is the largest of its kind. Furthermore, DTR is created from the samples of tweets in self-disclosed depressed users Twitter timeline from two datasets, including one of the largest benchmark datasets for user-level depression detection from Twitter. This further helps preserve the depression symptoms distribution of self-disclosed tweets. Subsequently, we iteratively retrain our initial DSD model with the harvested data. We discuss the stopping criteria and limitations of this SSL process, and elaborate the underlying constructs which play a vital role in the overall SSL process. We show that we can produce a final dataset which is the largest of its kind. Furthermore, a DSD and a Depression Post Detection model trained on it achieves significantly better accuracy than their initial version.

No abstract available

Automated depression detection (ADD) from speech signals allows early identification and intervention, reducing costs to medical healthcare. However, most of the existing ADD studies are trained and evaluated on a single language corpus with a lack of sufficient training data. These limits the generalizability of models in other demographic groups in distinct languages. In this study, Semi-Supervised Learning (SSL) was applied to depression detection on two different language datasets. We evaluate the HuBERT and WavLM models in single-language, mixed-language, and cross-language scenarios to investigate the generalization to diverse populations at different recording environments. Moreover, we thoroughly analyzed layer-wise performance in the upstream model and pooling methods (i.e. max and mean pooling) in the downstream task. The results show that the WavLM features generalize better than the HuBERT features. Our best model surpasses previous works in the frozen upstream conditions.

Depression is a widespread and severe mental health disorder, characterized by a significant prevalence and a high incidence rate, which presents numerous challenges in diagnosis and treatment. We introduce an innovative multimodal approach for depression recognition utilizing self-supervised learning. The method analyzes audio and video data through three steps: extracting image features, extracting audio features, and performing multimodal temporal fusion, enabling an assessment of depression severity. Extensive experiments conducted on the AVEC2014 dataset demonstrate that the proposed model achieves a Mean Absolute Error (MAE) of 7.013 and a Root Mean Square Error (RMSE) of 8.616 in depression prediction tasks, showcasing strong performance.

One of the most prevalent mental illnesses and a significant contributing factor to suicides is depression. The increased usage of social media platforms has made it possible to diagnose depression early on by speaking to individuals in their native tongue. This research explores sentiment analysis on a Twitter dataset by categorizing tweets into ‘positive’ and ‘negative’ sentiments. The proposed model exhibits exceptional performance in the context of depression and anxiety sentiment classification. Employing pre-processing techniques enhanced dataset quality, including tokenization, lemmatization, and stop word removal. The entire process, including the analysis and sorting steps, would use social media data, including text data and emoticons, emojis, and other graphics. Incorporating the TF-IDF weighting scheme further refined sentiment conveyance by emphasizing specific terms. A supervised machine-learning algorithm was applied, and the proposed model demonstrated superior performance, achieving precision, recall, F1-Score, and accuracy at 98.27%, 98.07%, 98.00%, and 98.01%, respectively. These remarkable metrics underscore the proposed model’s exceptional ability to classify tweets accurately, highlighting its robustness in handling X (formally Twitter) data distinctions for advanced social media analytics.

Automated Depression Detection (ADD) in speech aims to automatically estimate one's depressive attributes through artificial intelligence tools towards spoken signals. Nevertheless, existing speech-based ADD works fail to sufficiently consider weakly-supervised cases with inaccurate labels, which may typically appear in intelligent mental health. In this regard, we propose the Self-Learning-based Label Correction (SLLC) approach for weakly-supervised depression detection in speech. The proposed approach employs a self-learning manner connecting a label correction module and a depression detection module. Within the approach, the label correction module fuses likelihood-ratio-based and prototype-based label correction strategies in order to effectively correct the inaccurate labels, while the depression detection module aims at detecting depressed samples through a 1D convolutional recurrent neural network with multiple types of losses. The experimental results on two depression detection corpora show that our proposed SLLC approach performs better compared with existing state-of-the-art speech-based depression detection approaches, in the case of weak supervision with inaccurate labels for depression detection in speech.

Speech-based depression detection (SDD) has emerged as a non-invasive and scalable alternative to conventional clinical assessments. However, existing methods still struggle to capture robust depression-related speech characteristics, which are sparse and heterogeneous. Although pretrained self-supervised learning (SSL) models provide rich representations, most recent SDD studies extract features from a single layer of the pretrained SSL model for the downstream classifier. This practice overlooks the complementary roles of low-level acoustic features and high-level semantic information inherently encoded in different SSL model layers. To explicitly model interactions between acoustic and semantic representations within an utterance, we propose a hierarchical adaptive representation encoder with prior knowledge that disengages and re-aligns acoustic and semantic information through asymmetric cross-attention, enabling fine-grained acoustic patterns to be interpreted in semantic context. In addition, a Connectionist Temporal Classification (CTC) objective is applied as auxiliary supervision to handle the irregular temporal distribution of depressive characteristics without requiring frame-level annotations. Experiments on DAIC-WOZ and MODMA demonstrate that HAREN-CTC consistently outperforms existing methods under both performance upper-bound evaluation and generalization evaluation settings, achieving Macro F1 scores of 0.81 and 0.82 respectively in upper-bound evaluation, and maintaining superior performance with statistically significant improvements in precision and AUC under rigorous cross-validation. These findings suggest that modeling hierarchical acoustic-semantic interactions better reflects how depressive characteristics manifest in natural speech, enabling scalable and objective depression assessment.

Mood disorders such as depression and bipolar disorder are serious mental health problems that significantly affect patients’ quality of life. Traditional diagnostic processes, which rely on subjective interviews and clinical observations, are often insufficient to accurately distinguish disorders with overlapping symptoms accurately. Therefore, this study aims to develop and compare the performance of three supervised machine learning algorithms namely Logistic Regression, Random Forest, and Support Vector Machine (SVM)—in multi-class classification of four categories of mood disorders: Bipolar Type I, Bipolar Type II, Depression, and Normal. The dataset consists of 120 psychology patients with 17 key clinical symptoms used as input features. Data were split into training (60%), validation (20%), and testing (20%) sets, with hyperparameter tuning performed using GridSearchCV. The evaluation results show that Random Forest and Logistic Regression achieved the highest accuracy of 83.33%, while SVM achieved 79.17%. ROC Curve and AUC analysis confirmed that Random Forest has the best discriminatory power, with AUC values close to or equal to 1.00 across all classes. Furthermore, feature importance analysis revealed that Mood Swing, Optimism, and Sexual Activity were the most influential features in classification, highlighting their clinical relevance in distinguishing mood disorder types. This study demonstrates that supervised machine learning, particularly the Random Forest model, not only improves diagnostic accuracy but also provides insights into the most critical features, making it a valuable tool to support more objective clinical decision-making in mental health.

This document Among all the various types of mental and psychosocial illnesses, the most commonly occurring type is depression. It can cause serious problems such as suicide. Therefore, early detection is important to stop the progression of this disease that could endanger human lives. Predicting and detecting early-stage depression using machine learning (ML) techniques is a promising strategy. This study’s main purpose is to assess which ML techniques are highly appropriate and accurate regarding such diagnoses. Six supervised ML techniques namely: K-nearest neighbor (KNN), Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), Support vector machine (SVM) and Naive Bayes (NB) were applied on dataset collected from Kaggle and compared for their accuracy (ACC) and performance in predicting depression. The performance of each model was evaluated using 10-fold cross-validation and evaluated in terms of ACC, F1-score, Precision (PR), and Sensitivity (SEN). Based on the experimental results analysis, we can conclude that SVM and LR performed better than all other methods with an ACC of 83,32%. Therefore, we found that a simple ML algorithm can be used to assist clinicians and practitioners predict depression at an early stage, with excellent potential utility and a considerable degree of ACC.

Background: Depression in children and adolescents poses significant public health challenges, impacting individual and societal well-being. Traditional methods for diagnosing and predicting depression often rely on symptomatic assessments post-onset, delaying intervention. Recent advancements in machine learning (ML) provide new opportunities for early prediction and prevention. Methods: This study leveraged data from the U.S. National Health Interview Survey(NHIS) from 2004 to 2014, involving 27,642 participants aged 4-17. We employed machine learning models, including XGBoost, Random Forest, Decision Tree, and Bagging Classifier, to analyze a vast array of physical activity and health behavior data. These models were evaluated for their accuracy in predicting mental health indicators. Results: The XGBoost model demonstrated notable accuracy (77.5%) in forecasting mental health scores, closely followed by the Random Forest classifier (77.4%). The models were further assessed for their ability to classify varying degrees of mental anxiety, with the redefined categories of 'Mild', 'Moderate', and 'Severe'. The retrained XGBoost model showed enhanced accuracy across these categories (81.7%- 85.7%), with AUC values indicating reliable differentiation between mental health states. Discussion: This study expands the scope of ML in predicting depression, highlighting the intricate relationship between diverse health behaviors and mental health. The predictive accuracy of the models underscores the potential of M Linearly detection and intervention for depression. Future research should focus on refining these models for broader application and exploring their utility in real-world clinical settings.

A fundamental component of user-level social media language based clinical depression modelling is depression symptoms detection (DSD). Unfortunately, there does not exist any DSD dataset that reflects both the clinical insights and the distribution of depression symptoms from the samples of self-disclosed depressed population. In our work, we describe a Semi-supervised Learning (SSL) framework which uses an initial supervised learning model that leverages 1) a state-of-the-art large mental health forum text pre-trained language model further fine-tuned on a clinician annotated DSD dataset, 2) a Zero-Shot learning model for DSD, and couples them together to harvest depression symptoms related samples from our large self-curated Depression Tweets Repository (DTR). Our clinician annotated dataset is the largest of its kind. Furthermore, DTR is created from the samples of tweets in self-disclosed depressed users Twitter timeline from two datasets, including one of the largest benchmark datasets for user-level depression detection from Twitter. This further helps preserve the depression symptoms distribution of self-disclosed Twitter users tweets. Subsequently, we iteratively retrain our initial DSD model with the harvested data. We discuss the stopping criteria and limitations of this SSL process, and elaborate the underlying constructs which play a vital role in the overall SSL process. We show that we can produce a final dataset which is the largest of its kind. Furthermore, a DSD and a Depression Post Detection (DPD) model trained on it achieves significantly better accuracy than their initial version.

Introduction Globally, the prevalence of mental health problems, especially depression, is at an all-time high. The objective of this study is to utilize machine learning models and sentiment analysis techniques to predict the level of depression earlier in social media users' posts. Methods The datasets used in this research were obtained from Twitter posts. Four machine learning models, namely extreme gradient boost (XGB) Classifier, Random Forest, Logistic Regression, and support vector machine (SVM), were employed for the prediction task. Results The SVM and Logistic Regression models yielded the most accurate results when applied to the provided datasets. However, the Logistic Regression model exhibited a slightly higher level of accuracy compared to SVM. Importantly, the logistic regression model demonstrated the advantage of requiring less execution time. Discussion The findings of this study highlight the potential of utilizing machine learning models and sentiment analysis techniques for early detection of depression in social media users. The effectiveness of SVM and Logistic Regression models, with Logistic Regression being more efficient in terms of execution time, suggests their suitability for practical implementation in real-world scenarios.

Social media has seen enormous growth in its users recently and knowingly or unknowingly the behavior of a person will be reflected in the comments she/he posts on social media. Users having the sign of depression may post negative or disturbing content seeking the attention of other users. Hence, social media data can be analysed to check whether the users’ have the sign of depression and help them to get through the situation if required. However, as analyzing the increasing amount of social media data manually in laborious and error-prone, automated tools have to be developed for the same. To address the issue of detecting the sign of depression content on social media, in this paper, we - team MUCS, describe an Ensemble of Machine Learning (ML) models and a Transfer Learning (TL) model submitted to “Detecting Signs of Depression from Social Media Text-LT-EDI@ACL 2022” (DepSign-LT-EDI@ACL-2022) shared task at Association for Computational Linguistics (ACL) 2022. Both frequency and text based features are used to train an Ensemble model and Bidirectional Encoder Representations from Transformers (BERT) fine-tuned with raw text is used to train the TL model. Among the two models, the TL model performed better with a macro averaged F-score of 0.479 and placed 18th rank in the shared task. The code to reproduce the proposed models is available in github page1.

This study aims to enhance the accuracy of depression detection by leveraging representation learning from audio data. The data of depression speech sets are sparse and costly to annotate. Therefore, a self-supervised pre-training approach is employed to improve the performance, generalization capability, and training efficiency of downstream tasks. When processing unlabeled data, the pre-trained audio representations based on self-supervised learning may be interfered with by noisy data if there is a significant amount of noise or errors present. Consequently, it is necessary to effectively analyze long-distance sequence data to enhance anti-interference capabilities. However, traditional LSTM models have limitations in context extraction and robustness to input outliers. Thus, an improved method named CNN-BiLSTM is proposed in this paper. The network initializes the LSTM's embedding layer with pre-trained word vectors and extracts spatial and temporal features separately to ensure a full and complete expression of useful input information. Different weights are assigned based on the importance of the features to obtain fused features. Additionally, a random forest is used for classification to mitigate the risk of overfitting and to demonstrate good performance when processing high-dimensional data. Experimental results show that the proposed model exhibits good classification performance on the depression dataset, outperforming traditional methods and state-of-the-art investigations.

No abstract available

Although social anxiety and depression are common, they are often underdiagnosed and undertreated, in part due to difficulties identifying and accessing individuals in need of services. Current assessments rely on client self-report and clinician judgment, which are vulnerable to social desirability and other subjective biases. Identifying objective, nonburdensome markers of these mental health problems, such as features of speech, could help advance assessment, prevention, and treatment approaches. Prior research examining speech detection methods has focused on fully supervised learning approaches employing strongly labeled data. However, strong labeling of individuals high in symptoms or state affect in speech audio data is impractical, in part because it is not possible to identify with high confidence which regions of a long speech indicate the person's symptoms or affective state. We propose a weakly supervised learning framework for detecting social anxiety and depression from long audio clips. Specifically, we present a novel feature modeling technique named NN2Vec that identifies and exploits the inherent relationship between speakers' vocal states and symptoms/affective states. Detecting speakers high in social anxiety or depression symptoms using NN2Vec features achieves F-1 scores 17% and 13% higher than those of the best available baselines. In addition, we present a new multiple instance learning adaptation of a BLSTM classifier, named BLSTM-MIL. Our novel framework of using NN2Vec features with the BLSTM-MIL classifier achieves F-1 scores of 90.1% and 85.44% in detecting speakers high in social anxiety and depression symptoms.

There is a significant correlation between depression, verbal behavior, and facial expressions. By analyzing patients’ audio and facial visuals, depression assessments can be conducted. However, existing work is predominantly based on single modalities. Additionally, acquiring a sufficient amount of labeled data in clinical settings is challenging and costly. To leverage multimodal audio-visual data while addressing the issue of lacking trainable labeled data, we propose an audiovisual multimodal semi-supervised depression detection model based on Graph Neural Networks (AVS-GNN). This model first extracts dual-modality temporal information from audio features and facial visual features of patients and obtains modality-specific high-level embedding representations through graph representation learning. Subsequently, it utilizes graph-based contrastive unsupervised learning to capture consistency information between pairs of unlabeled samples across different modalities and to facilitate cross-modal interactions. We specifically designed a hybrid weighted pseudo-labeling strategy to assign high-confidence pseudo-labels to unlabeled data and further retraining the model. Experiments on two depression datasets show that this model outperforms baseline methods across all evaluation metrics.

Background Passive sensor data from mobile devices can shed light on daily activities, social behavior, and maternal-child interactions to improve maternal and child health services including mental healthcare. We assessed feasibility and acceptability of the Sensing Technologies for Maternal Depression Treatment in Low Resource Settings (StandStrong) platform. The StandStrong passive data collection platform was piloted with adolescent and young mothers, including mothers experiencing postpartum depression, in Nepal. Methods Mothers (15–25 years old) with infants (< 12 months old) were recruited in person from vaccination clinics in rural Nepal. They were provided with an Android smartphone and a Bluetooth beacon to collect data in four domains: the mother’s location using the Global Positioning System (GPS), physical activity using the phone’s accelerometer, auditory environment using episodic audio recording on the phone, and mother-infant proximity measured with the Bluetooth beacon attached to the infant’s clothing. Feasibility and acceptability were evaluated based on the amount of passive sensing data collected compared to the total amount that could be collected in a 2-week period. Endline qualitative interviews were conducted to understand mothers’ experiences and perceptions of passive data collection. Results Of the 782 women approached, 320 met eligibility criteria and 38 mothers (11 depressed, 27 non-depressed) were enrolled. 38 mothers (11 depressed, 27 non-depressed) were enrolled. Across all participants, 5,579 of the hour-long data collection windows had at least one audio recording [mean ( M ) = 57.4% of the total possible hour-long recording windows per participant; median ( Mdn ) = 62.6%], 5,001 activity readings ( M  = 50.6%; Mdn  = 63.2%), 4,168 proximity readings ( M  = 41.1%; Mdn  = 47.6%), and 3,482 GPS readings ( M  = 35.4%; Mdn  = 39.2%). Feasibility challenges were phone battery charging, data usage exceeding prepaid limits, and burden of carrying mobile phones. Acceptability challenges were privacy concerns and lack of family involvement. Overall, families’ understanding of passive sensing and families’ awareness of potential benefits to mothers and infants were the major modifiable factors increasing acceptability and reducing gaps in data collection. Conclusion Per sensor type, approximately half of the hour-long collection windows had at least one reading. Feasibility challenges for passive sensing on mobile devices can be addressed by providing alternative phone charging options, reverse billing for the app, and replacing mobile phones with smartwatches. Enhancing acceptability will require greater family involvement and improved communication regarding benefits of passive sensing for psychological interventions and other health services. Registration International Registered Report Identifier (IRRID): DERR1-10.2196/14734

No abstract available

Depression is a significant mental health issue that profoundly impacts people’s lives. Diagnosing depression often involves interviews with mental health professionals and surveys, which can become cumbersome when administered continuously. Digital phenotyping offers an innovative approach for detecting and monitoring depression without requiring active user involvement. This study contributes to the detection of depression severity and depressive symptoms using mobile devices. Our proposed approach aims to distinguish between different patterns of depression and improve prediction accuracy. We conducted an experiment involving 381 participants over a period of at least three months, during which we collected comprehensive passive sensor data and Patient Health Questionnaire (PHQ-9) self-reports. To enhance the accuracy of predicting depression severity levels (classified as none/mild, moderate, or severe), we introduce a novel approach called symptom profiling. The symptom profile vector represents nine depressive symptoms and indicates both the probability of each symptom being present and its significance for an individual. We evaluated the effectiveness of the symptom-profiling method by comparing the F1 score achieved using sensor data features as inputs to machine learning models with the F1 score obtained using the symptom profile vectors as inputs. Our findings demonstrate that symptom profiling improves the F1 score by up to 0.09, with an average improvement of 0.05, resulting in a depression severity prediction with an F1 score as high as 0.86.

No abstract available

It has recently been reported that identifying the depression severity of a person requires involvement of mental health professionals who use traditional methods like interviews and self-reports, which results in spending time and money. In this work we made solid contributions on short-term depression detection using every-day mobile devices. To improve the accuracy of depression detection, we extracted five factors influencing depression (symptom clusters) from the DSM-5 (Diagnostic and Statistical Manual of Mental Disorders), namely, physical activity, mood, social activity, sleep, and food intake and extracted features related to each symptom cluster from mobile devices’ sensors. We conducted an experiment, where we recruited 20 participants from four different depression groups based on PHQ-9 (the Patient Health Questionnaire-9, the 9-item depression module from the full PHQ), which are normal, mildly depressed, moderately depressed, and severely depressed and built a machine learning model for automatic classification of depression category in a short period of time. To achieve the aim of short-term depression classification, we developed Short-Term Depression Detector (STDD), a framework that consisted of a smartphone and a wearable device that constantly reported the metrics (sensor data and self-reports) to perform depression group classification. The result of this pilot study revealed high correlations between participants` Ecological Momentary Assessment (EMA) self-reports and passive sensing (sensor data) in physical activity, mood, and sleep levels; STDD demonstrated the feasibility of group classification with an accuracy of 96.00% (standard deviation (SD) = 2.76).

Background Passive mobile sensing provides opportunities for measuring and monitoring health status in the wild and outside of clinics. However, longitudinal, multimodal mobile sensor data can be small, noisy, and incomplete. This makes processing, modeling, and prediction of these data challenging. The small size of the data set restricts it from being modeled using complex deep learning networks. The current state of the art (SOTA) tackles small sensor data sets following a singular modeling paradigm based on traditional machine learning (ML) algorithms. These opt for either a user-agnostic modeling approach, making the model susceptible to a larger degree of noise, or a personalized approach, where training on individual data alludes to a more limited data set, giving rise to overfitting, therefore, ultimately, having to seek a trade-off by choosing 1 of the 2 modeling approaches to reach predictions. Objective The objective of this study was to filter, rank, and output the best predictions for small, multimodal, longitudinal sensor data using a framework that is designed to tackle data sets that are limited in size (particularly targeting health studies that use passive multimodal sensors) and that combines both user agnostic and personalized approaches, along with a combination of ranking strategies to filter predictions. Methods In this paper, we introduced a novel ranking framework for longitudinal multimodal sensors (FLMS) to address challenges encountered in health studies involving passive multimodal sensors. Using the FLMS, we (1) built a tensor-based aggregation and ranking strategy for final interpretation, (2) processed various combinations of sensor fusions, and (3) balanced user-agnostic and personalized modeling approaches with appropriate cross-validation strategies. The performance of the FLMS was validated with the help of a real data set of adolescents diagnosed with major depressive disorder for the prediction of change in depression in the adolescent participants. Results Predictions output by the proposed FLMS achieved a 7% increase in accuracy and a 13% increase in recall for the real data set. Experiments with existing SOTA ML algorithms showed an 11% increase in accuracy for the depression data set and how overfitting and sparsity were handled. Conclusions The FLMS aims to fill the gap that currently exists when modeling passive sensor data with a small number of data points. It achieves this through leveraging both user-agnostic and personalized modeling techniques in tandem with an effective ranking strategy to filter predictions.

Anxiety disorders are the most common class of psychiatric problems affecting both children and adults. However, tools to effectively monitor and manage anxiety are lacking, and comparatively limited research has been applied to addressing the unique challenges around anxiety. Leveraging passive and unobtrusive data collection from smartphones could be a viable alternative to classical methods, allowing for real-time mental health surveillance and disease management. This paper presents eWellness, an experimental mobile application designed to track a full-suite of sensor and user-log data off an individual's device in a continuous and passive manner. We report on an initial pilot study tracking ten people over the course of a month that showed a nearly 76% success rate at predicting daily anxiety and depression levels based solely on the passively monitored features.

Mental health disorders affect approximately 1 billion people worldwide, yet early detection remains challenging due to limited access to healthcare resources and stigma associated with seeking help. This paper presents DeepMindWatch, a novel multimodal artificial intelligence framework for early detection of depression and anxiety disorders through analysis of digital biomarkers. The framework employs a hierarchical attention network that processes these heterogeneous data streams and generates interpretable predictions with supporting evidence. We trained our models on a dataset combining anonymized clinical records (n = 1,850), consented social media data (n = 5,200), and passive sensing data (n = 3,400) collected through our custom smartphone application. The proposed framework achieves 87.3% accuracy in identifying depression markers and 82.6% accuracy for anxiety markers, outperforming existing unimodal approaches by 14.2% on average DeepMindWatch demonstrates robust performance across diverse demographic groups with a reduced false positive rate of 8.4% compared to clinical screening tools. A six-month longitudinal validation study confirms the framework's efficacy in early detection, identifying potential mental health concerns an average of 63 days before clinical diagnosis. This research offers promising directions for developing scalable, privacy-preserving digital mental health screening tools that could significantly improve early intervention strategies and accessibility of mental healthcare resources globally.

Given the prevalence of missing data in longitudinal passive sensing studies, data imputation -- a critical preprocessing step -- is often overlooked by researchers in favor of other aspects of data analyses, like building sophisticated models or outcome prediction. In this paper, we seek to direct the attention of the behavioral and mental health-sensing community toward the importance of data imputation in such studies. In this work, we evaluate and benchmark off-the-shelf imputation strategies using the open-source GLOBEM platform and datasets. Our results demonstrate that using appropriate imputation strategies could improve performance by up to 25% increase in AUROC for predicting participants? future depression labels (self-reported PHQ-4) using past sensing data with the same model building and prediction pipeline as the GLOBEM platform, without compromising the inherent underlying structure of behavioral sensing data post-imputation. Furthermore, we observe that certain imputation strategies significantly improve the separability of predicted depression probabilities on the test data, compared to no or trivial imputation. Lastly, we present a case study of users with changing depression labels and demonstrate that by using these imputation strategies, we are better able to capture and trace within-person transitions of depression as compared to trivial or no imputation.

Longitudinal passive sensing studies for health and behavior outcomes often have missing and incomplete data. Handling missing data effectively is thus a critical data processing and modeling step. Our formative interviews with researchers working in longitudinal health and behavior passive sensing revealed a recurring theme: most researchers consider imputation a low-priority step in their analysis and inference pipeline, opting to use simple and off-the-shelf imputation strategies without comprehensively evaluating its impact on study outcomes. Through this paper, we call attention to the importance of imputation. Using publicly available passive sensing datasets for depression, we show that prioritizing imputation can significantly impact the study outcomes - with our proposed imputation strategies resulting in up to 31% improvement in AUROC to predict depression over the original imputation strategy. We conclude by discussing the challenges and opportunities with effective imputation in longitudinal sensing studies.

Self‐report questionnaires, used for detecting major depressive disorder (MDD) in daily life, may incur biases stemming from social desirability and repetitive answers. Though detection based on mobile sensing was being developed recently, it cannot sufficiently promote self‐help action due to the characteristics of passive feedback. Thus, an active self‐monitoring and feedback system is crucial for individuals to recognize and address their malfunctions. In this study, we proposed to predict changes in MDD severity using cognitive tasks monitored on mobile devices. An online survey was conducted to evaluate the severity, incorporating cognitive tasks such as Navon task, Go/No‐go task, and n‐back task, along with the Quick Inventory of Depressive Symptomatology. Participants completed the survey three times on their mobile devices. The analysis included data from 75 participants, including 21 participants whose MDD score increased by at least one point during the second and third surveys; the first survey was excluded to avoid confounding effects. A random forest classifier was employed for classifying participants whose depression has and has not worsened. The learned model achieved modest accuracy (68.3%) with a significant mean area under the curve of 0.59 (t(9) = 2.98, p = .016, dz = 0.94), suggesting the potential to predict depressive states based on cognitive domains. Moreover, working memory and attentional inhibition functions contributed to predicting the severity change mostly. Though improvements are required to reduce false negatives for practical applications, our result suggests that MDD aggravation could be assessed by mobile cognitive tasks.

Smartphones are widely used as portable data collectors for wearable and healthcare sensors that can passively collect data streams related to the environment, health status, and behaviors. Recent research shows that the collected data can be used to monitor not only the physical states but also the mental health of individuals. However, extracting the features of digital phenotypes that characterize major depressive disorder (MDD) is technically challenging and may raise significant privacy concerns. Addressing such challenges has become the focus of many researchers. This article provides a comprehensive analysis of several key issues related to ubiquitous sensing to aid in detecting MDD. Specifically, this article analyzes existing methodologies and feature extraction algorithms used to detect possible MDD through digital phenotyping from smartphone data. In particular, five types of features are summarized and explained, namely, location, movement, rhythm, sleep, and social and device usage. Finally, related limitations and challenges are discussed to provide paths for further research and engineering.

In this work, we propose a combined sampling technique to improve the performance of imbalanced classification of university student depression data. In experimental results, we found that combined random oversampling with the Tomek links under sampling methods allowed generating a relatively balanced depression dataset without losing significant information. In this case, the random oversampling technique was used for sampling the minority class to balance the number of samples between the datasets. Then, the Tomek links technique was used for undersampling the samples by removing the depression data considered less relevant and noisy. The relatively balanced dataset was classified by random forest. The results show that the overall accuracy in the prediction of adolescent depression data was 94.17%, outperforming the individual sampling technique. Moreover, our proposed method was tested with another dataset for its external validity. This dataset’s predictive accuracy was found to be 93.33%.

Adolescent mental health problems are frequent, and traditional intervention methods have limitations in precision and personalization. This study constructs an intelligent intervention framework driven by algorithms and deep learning, which enhances the scientificity and accessibility of adolescent mental health services through multimodal data fusion and dynamic strategy optimization. The research design includes a closed-loop system of "data preprocessing dynamic evaluation intelligent intervention effectiveness feedback". The dynamic mental health assessment model integrates multiple sources of data such as psychological scales, physiological signals, and social texts, and adopts a multi task learning architecture of random forest and long short-term memory network (LSTM) to construct a three-level feature system to capture the static features and dynamic evolution of psychological states. The experiment showed that the AUC-ROC for predicting depression and anxiety reached 0.89 and 0.87, respectively, which was 17% higher than traditional scale evaluation. The misdiagnosis rate of subclinical symptoms was reduced by 60%, verifying the efficient representation ability of multimodal data for complex psychological characteristics. The innovative paradigm of "data intelligence+psychological theory" has been provided for the intervention of adolescent mental health, and its technical framework and empirical results have important reference value for the construction of intelligent psychological service system.

Adolescent suicide is a critical public health issue, yet accurately predicting suicide risk remains challenging. Few studies integrate adolescents' self-reports with mental health, especially suicidality assessments from parents and siblings. This study employed machine learning (ML) models - Random Forest, XGBoost and LightGBM - to analyze 169 demographic, socio-relational, mental health, and functional variables from 7286 adolescents and one of their parents and siblings (N = 21,858) to identify key predictors of suicide risk. There were 14.8 % of the adolescents at risk. Family-triad-models, which used all data from target adolescent, their sibling and parent, achieved consistent patterns in variable importance for predicting suicidality, with AUROC scores ranging from 0.875 to 0.877. All nine top predictors identified by all three ML methods originated from adolescents' self-reports, with emotional difficulties as the most important predictor, adjusted odds ratio (aOR) per ±1SD being 1.59 [95 % CI 1.46-1.73]. In family-proxy-models, which excluded the 56 self-reported variables from the target adolescent, predictive accuracy declined but remained sufficient (AUROC: 0.621 to 0.625). Four variables were identified by all three ML methods, with parents' reports of the target adolescent's emotional difficulties being the strongest, aOR ± 1SD being 1.47 [95 % CI 1.39-1.56]. Machine learning can effectively leverage adolescents' self-reports to predict suicide risk accurately. Even when adolescents' self-reports are unavailable due to their unwillingness to disclose information, family members' reports alone provide a sufficiently accurate basis for prediction. Moreover, emotional difficulties perceived by both adolescents and parents are crucial indicators of suicidality.

This paper presents a system aimed at preventing the serious social issue of adolescent suicide by identifying at-risk adolescents early and providing appropriate support. To achieve this, we propose machine learning models based on data from the Korea Youth Risk Behavior Web-Based Survey (KYRBWS), targeting South Korean adolescents and considering various factors to predict suicide attempts. The models incorporate variables such as adolescents' behavioral patterns, emotional states, and family and social support systems. We applied Random Forest for feature selection to identify important features from the original dataset. The core algorithms include Logistic Regression, Random Forest, and XGBoost. To ensure robust performance, we employed K-fold cross-validation. Additionally, to address the challenge of data imbalance, we applied various sampling techniques to develop reliable, high-performance models for predicting suicide attempts. By detecting adolescents at high risk of suicide attempts, we aim to contribute to suicide prevention efforts and promote adolescent health and well-being.

Background: Parent-child conflict is a known risk factor for adolescent suicidal ideation but remains underexplored using predictive models. Objectives: This study aimed to predict adolescent suicidal ideation from parent-child conflict using machine learning. Methods: The study involved 442 adolescent girls from Tehran province selected via convenience sampling. Data were collected using the Beck Suicidal Ideation Scale (BSSI) and the Parent-Child Behavioral Conflict Questionnaire (CBQ). Analysis was conducted using Pearson correlation and four machine learning algorithms: Logistic regression, SVM, random forest, and XGBoost. Results: The findings revealed that certain aspects of parent-child conflict, such as incompatibility, feeling misunderstood, and shouting, were significantly associated with the severity of suicidal ideation, with correlation coefficients reaching up to 0.259. Machine learning models, particularly SVM and random forest, predicted suicidal ideation risk with an accuracy of 78.79%, while XGBoost showed a lower accuracy of 69.70%. Conclusions: This study emphasizes the role of family relationship quality in adolescent suicide prevention and supports using intelligent models for early screening.

No abstract available

The pathogenesis of depression is highly complex, therefore, the development of predictive models using readily available clinical parameters to identify individuals at risk of adverse depressive outcomes holds significant clinical value. 7108 participants from the United States National Health and Nutrition Examination Survey were collected. A total of 11 machine learning models were employed, including CatBoost, Decision Tree, Gradient Boosting Tree, LightGBM (LGB), Logistic Regression (LR), Lasso, Naive Bayes, Neural Network, Random Forest (RF), Support Vector Machine, and XGBoost, with comparisons made against the generalized linear regression model. Model performance was rigorously assessed using receiver operating characteristic (ROCs), calibration curves, and decision curves analysis. Feature importance was interpreted through Shapley Additive exPlanations to identify key influencing factors at the whole level and interpret individual heterogeneity through instance-level analysis. Significant differences in overall characteristics were observed between depressed patients and healthy controls. The RF model demonstrated superior performance, followed by Lasso, XGBoost, and LGB models, which also showed relatively high predictive accuracy. The training set AUC values for the RF, Lasso, XGBoost, and LGB models were 0.998, 0.713, 0.723, and 0.804, respectively, while their corresponding test set AUC values were 0.705, 0.719, 0.714, and 0.687. Based on variable importance ranking from RF, Lasso, XGBoost, and LGB models, we identified eight key predictors: body mass index, education level, marital status, annual family income, family income to poverty ratio, trouble sleeping, composite dietary antioxidant index, and dietary inflammatory index. These variables were integrated to develop a comprehensive statistical model for predicting depression risk. We developed a robust predictive model for assessing depression risk, incorporating eight clinically accessible predictors. This model demonstrates reliable predictive performance for depression onset and provides valuable reference for clinical decision-making. Clinical trial number is not applicable.

The rising prevalence of depression among students has drawn significant attention. As data mining and artificial intelligence technologies continue to evolve, leveraging behavioral and textual data for early depression prediction offers new opportunities for timely mental health interventions. This study uses 17 depression-related features, including gender, financial stress, academic pressure, and working hours. To evaluate model performance, this study focused on three representative classification algorithms. The first is logistic regression, known for its interpretability; the second is random forest, which leverages ensemble learning; and the third is XGBoost, a powerful gradient boosting framework. To assess model performance, this study considered multiple quantitative indicators. These include the proportion of correct predictions (accuracy), the model’s ability to identify true positives (precision and recall), the harmonic means of those two (F1 score), and the area under the ROC curve (AUC), which reflects the overall classification capability. A 50-fold robustness test was also conducted to validate model stability. SHAP plots were utilized to interpret model predictions and to identify the most influential features contributing to depression risk at the end of this paper. The experimental data showed that Logistic Regression had the highest AUC score, which was 0.913, while the AUC scores of Random Forest and XGBoost were 0.906 and 0.903, respectively. Calibration curve analysis verified that Logistic Regression had the best calibration performance. The result of this study supports the feasibility and value of Logistic Regression in predicting student depression.

Objective To explore the risk factors that affect adolescents’ suicidal and self-injurious behaviors and to construct a prediction model for adolescents’ suicidal and self-injurious behaviors based on machine learning algorithms. Methods Stratified cluster sampling was used to select high school students in Chongqing, yielding 3,000 valid questionnaires. Based on whether students had engaged in suicide or self-injury, they were categorized into a suicide/self-injury group (n=78) and a non-suicide/self-injury group (n=2,922). Gender, age, insomnia, and mental illness data were compared between the two groups, and a logistic regression model was used to analyze independent risk factors for adolescent suicidal and self-injurious behavior. Six methods—multi-level perceptron, random forest, K-nearest neighbor, support vector machine, logistic regression, and extreme gradient boosting—were used to build predictive models. Various model indicators for suicidal and self-injurious behavior were compared across the six algorithms using a confusion matrix to identify the optimal model. Result In the self-injury and suicide groups, the proportions of male adolescents, late adolescence, insomnia, and mental illness were significantly higher than in the non-suicide and self-injury groups (p <0.05). Compared with the non-suicidal self-injury group, this group also showed significantly increased scores in cognitive subscales, impulsivity, psychoticism, introversion–extroversion, neuroticism, interpersonal sensitivity, depression, anxiety, hostility, terror, and paranoia (p <0.05). These statistically significant variables were analyzed in a logistic regression model, revealing that gender, impulsivity, psychoticism, neuroticism, interpersonal sensitivity, depression, and paranoia are independent risk factors for adolescent suicide and self-injury. The logistic regression model achieved the highest sensitivity and specificity in predicting adolescent suicide and self-injury behavior (0.9948 and 0.9981, respectively). Performance of the random forest, multi-level perceptron, and extreme gradient models was acceptable, while the K-nearest neighbor algorithm and support vector machine performed poorly. Conclusion The detection rate of suicidal and self-injurious behaviors is higher in women than in men. Adolescents displaying impulsiveness, psychoticism, neuroticism, interpersonal sensitivity, depression, and paranoia have a greater likelihood of engaging in such behaviors. The machine learning model for classifying and predicting adolescent suicide and self-injury risk effectively identifies these behaviors, enabling targeted interventions.

No abstract available

: Depression is a growing mental health problem among young people in Brazil, with factors such as socioeconomic and lifestyle conditions influencing its prevalence. This study investigates how variables such as education, family situation, and access to services impact the incidence of depression, using data from the National Health Survey (PNS) of the Brazilian Institute of Geography and Statistics (IBGE). Using machine learning algorithms such as Random Forest, XGBoost, SVM, and MLP, the analysis identified patterns among the factors, highlighting sleep problems and depressive feelings as the main determinants, with Recall above 70%. These results support the creation of more inclusive mental health policies.

This study develops a systematic and interpretable framework for depression risk modeling using machine learning approaches. It utilizes a Kaggle dataset with demographic, academic, and psychological factors, applying several models including Logistic Regression, SVM, Random Forest, XGBoost, LightGBM, and MLP. The models were optimized using hyperparameter tuning and cross-validation. Among them, XGBoost achieved the highest performance, with an accuracy of 93.51% and an F1-score of 0.94 on the training set. The study also used Spearman's rank correlation to identify key variables influencing depression, such as suicidal thoughts (r ≈ 0.55) and family history of mental illness (r ≈ 0.36). Additionally, SHAP (SHapley Additive exPlanations) was employed for model interpretability, providing insights into the most influential features for depression prediction. These results offer actionable insights for early mental health screening and intervention.

Millions of people across the globe suffer from depression, a common mental illness that needs to be diagnosed and treated early to avoid this from turning into something much serious. This paper aims at designing a system that predicts the likelihood of depression based on the structured questionnaire responses with the use of machine learning-based approach. The dataset is balanced with respect to the target classes "Depressed" and "Not Depressed" and comprises answers to 14 questions. Accuracy, precision, recall, F1-score, and confusion matrices for five machine learning models were compared: Logistic Regression, Random Forest, XGBoost, Support Vector Machine (SVM), and an Artificial Neural Network (ANN). SVM performed the best in the contest of prediction accuracy and consistency by having an accuracy of 99% on testing.The proposed approach will prove to be a valuable asset for mental health professionals in allowing rapid and accurate depression screening. Future research will attempt to include larger datasets and explore more features in improving the robustness of prediction.

This work aims to predict depression based on diverse data using Machine Learning Algorithms. The designed model seeks to identify early indicators of depression, providing a potential tool for proactive intervention and support in mental health by analyzing patterns in behavioral, physiological, and contextual data. Machine learning algorithms, namely decision trees, extra trees, XGBoost, Stochastic gradient descent, grid search CV, Stacking, and Voting classifiers, etc., are used to predict depression in the early stage. This study emphasizes integrating machine learning techniques to enhance predictive accuracy and contribute to developing accessible and timely depression detection systems. The F1 score was added, which helped to identify the best machine learning algorithm among the ones applied. We have achieved an accuracy of 92 % with random forest, which is 3% higher than the work previously done in RF. We also achieved a 0.99 F1 score using Linear SVM.

Depression is one of the most common mental health disorders in the world, affecting millions of people. Early detection of depression is crucial for effective medical intervention. Multimodal networks can greatly assist in the detection of depression, especially in situations where in patients are not always aware of or able to express their symptoms. By analyzing text and audio data, such networks are able to automatically identify patterns in speech and behavior that indicate a depressive state. In this study, we propose two multimodal information fusion networks: early and late fusion. These networks were developed using convolutional neural network (CNN) layers to learn local patterns, a bidirectional LSTM (Bi-LSTM) to process sequences, and a self-attention mechanism to improve focus on key parts of the data. The DAIC-WOZ and EDAIC-WOZ datasets were used for the experiments. The experiments compared the precision, recall, f1-score, and accuracy metrics for the cases of using early and late multimodal data fusion and found that the early information fusion multimodal network achieved higher classification accuracy results. On the test dataset, this network achieved an f1-score of 0.79 and an overall classification accuracy of 0.86, indicating its effectiveness in detecting depression.

Introduction Depression is a serious mental health disease. Traditional scale-based depression diagnosis methods often have problems of strong subjectivity and high misdiagnosis rate, so it is particularly important to develop automatic diagnostic tools based on objective indicators. Methods This study proposes a deep learning method that fuses multimodal data to automatically diagnose depression using facial video and audio data. We use spatiotemporal attention module to enhance the extraction of visual features and combine the Graph Convolutional Network (GCN) and the Long and Short Term Memory (LSTM) to analyze the audio features. Through the multi-modal feature fusion, the model can effectively capture different feature patterns related to depression. Results We conduct extensive experiments on the publicly available clinical dataset, the Extended Distress Analysis Interview Corpus (E-DAIC). The experimental results show that we achieve robust accuracy on the E-DAIC dataset, with a Mean Absolute Error (MAE) of 3.51 in estimating PHQ-8 scores from recorded interviews. Discussion Compared with existing methods, our model shows excellent performance in multi-modal information fusion, which is suitable for early evaluation of depression.

This study investigates the potential of multimodal data integration, which combines electroencephalogram (EEG) data with sociodemographic characteristics like age, sex, education, and intelligence quotient (IQ), to diagnose mental diseases like schizophrenia, depression, and anxiety. Using Apache Spark and convolutional neural networks (CNNs), a data-driven classification pipeline has been developed for big data environment to effectively analyze massive datasets. In order to evaluate brain activity and connection patterns associated with mental disorders, EEG parameters such as power spectral density (PSD) and coherence are examined. The importance of coherence features is highlighted by comparative analysis, which shows significant improvement in classification accuracy and robustness. This study emphasizes the significance of holistic approaches for efficient diagnostic tools by integrating a variety of data sources. The findings open the door for creative, data-driven approaches to treating psychiatric diseases by demonstrating the potential of utilizing big data, sophisticated deep learning methods, and multimodal datasets to enhance the precision, usability, and comprehension of mental health diagnostics.

Objective Depression is a prevalent mental health disorder affecting millions of people. Traditional diagnostic methods primarily rely on self-reported questionnaires and clinical interviews, which can be subjective and vary significantly between individuals. This paper introduces the Integrative Multimodal Depression Detection Network (IMDD-Net), a novel deep-learning framework designed to enhance the accuracy of depression evaluation by leveraging both local and global features from video, audio, and text cues. Methods The IMDD-Net integrates these multimodal data streams using the Kronecker product for multimodal fusion, facilitating deep interactions between modalities. Within the audio modality, Mel Frequency Cepstrum Coefficient (MFCC) and extended Geneva Minimalistic Acoustic Parameter Set (eGeMAPS) features capture local and global acoustic properties, respectively. For video data, the TimeSformer network extracts both fine-grained and broad temporal features, while the text modality utilizes a pre-trained BERT model to obtain comprehensive contextual information. The IMDD-Net’s architecture effectively combines these diverse data types to provide a holistic analysis of depressive symptoms. Results Experimental results on the AVEC 2014 dataset demonstrate that the IMDD-Net achieves state-of-the-art performance in predicting Beck Depression Inventory-II (BDI-II) scores, with a Root Mean Square Error (RMSE) of 7.55 and a Mean Absolute Error (MAE) of 5.75. A classification to identify potential depression subjects can achieve an accuracy of 0.79. Conclusion These results underscore the robustness and precision of the IMDD-Net, highlighting the importance of integrating local and global features across multiple modalities for accurate depression prediction.

Effective intervention and the avoidance of long-term psychological and emotional repercussions depend on early recognition of depression. However, because its early symptoms are subtle, complex, and vary from person to person, they are frequently disregarded [1]. Timely diagnosis is made more difficult by the fact that many persons in the early stages of depression may not seek care or may find it difficult to express their feelings [2]. This study presents a novel artificial intelligence (AI) framework for analysing multimodal data, including text, voice tone, and facial expressions, in order to identify early indicators of depression. The proposed system integrates cutting-edge deep learning modals: BERT is used for understanding contextual linguistic cues [3], CNNs extract significant emotional indicators from facial features [4], and RNNs capture the temporal dynamics and tone shifts in speech [5]. These modalities are fused through a structured data integration strategy, enabling the system to interpret emotional patterns more holistically and accurately. When tested using benchmark datasets like DAIC-WOZ [6], the system shows excellent accuracy and dependability in real-time, non-intrusive identification of depressed signs. Deeper emotional analysis is made possible by the integration of language, auditory, and visual information, which also increases the model’s generalizability and robustness across a range of topics [7]. With its scalable, easily available, and objective tools that enhance conventional approaches, this work demonstrates the expanding potential of AI in mental health care [8]. This paradigm facilitates prompt diagnosis and creates opportunities for tailored intervention methods by providing professionals with early, data-driven insights. In the end, it brings us one step closer to a time when technology can help to improve mental health and lessen the prevalence of untreated depression worldwide.

Depression classification often relies on multimodal features, but existing models struggle to capture the similarity between multimodal features. Moreover, the social stigma surrounding depression leads to limited availability of datasets, which constrains model accuracy. This study aims to improve multimodal depression recognition methods by proposing a Multimodal Generation-Text Depression Classification Model. The model introduces a Multimodal-Deep-Extract-Feature Net to capture both long- and short-term sequential features. A Dual Text Contrastive Learning Module is employed to generate emotionally salient word embeddings from patients' transcribed text. Contrastive learning brings similar features closer and pushes dissimilar features apart, thereby enhancing the representation of dual-text features. Finally, a Joint Multi-modal Fusion Attention mechanism is proposed to amplify the representation of dominant modalities, effectively integrate all modalities, and capture global multimodal features. This integrated approach improves depression recognition accuracy, facilitating timely intervention and support for patients. The model achieves accuracy rates of 89.5% on the DAIC-Woz dataset and 92% on the MDD2024 dataset.

Depression is a mental disorder experienced by many people around the world. If not treated properly, this condition can have very serious consequences, even leading to suicide. According to WHO data, around 280 million people worldwide suffer from depression, and over 700,000 young people between the ages of 15 and 26 take their own lives each year. Signs of depression include a lack of social support, persistent feelings of sadness, and a loss of enthusiasm for daily activities. This study uses data in the form of text and images from tweets in Indonesian. To improve text representation, researchers used the FastText expansion feature, which is capable of replacing words with similar meanings. The amount of tweet data used was 24,235 and for IndoNews data it was 127,579. Meanwhile, image features were extracted using CNN EfficientNetB0 for more accurate analysis results.The data was then sorted using a Hybrid Deep Learning method that mixes Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM). The CNN was used to find important features from text and images, while BiLSTM helped understand the relationship between words in a sentence. Based on the experiment result, the CNN BiLSTM model demonstrated that a multimodal approach is capable of achieving the highest accuracy of 79.54% and precision of 82.14%. Compared to the baseline model CNN with accuracy of 75.44% and BiLSTM accuracy 73.09%, the proposed multimodal model improved the accuracy by 92.99%, confirming its significant contribution to performance enhancement p < 0.05. This proves that the combination of text and image data enriched with FastText expansion features can significantly improve the detection rate of depression. In addition, evaluation using a Confusion Matrix also shows that this model has stable and effective classification capabilities in distinguishing between users who are depressed and those who are not.

No abstract available

Automatic depression classification from multimodal input data is a challenging task. Modern methods use paralinguistic information such as audio and video signals. Using linguistic information such as speech signals and text data for depression classification is a complicated task in deep learning models. Best audio and video features are built to produce a dependable depression classification system. Textual signals related to depression classification are analyzed using text-based content data. Moreover, to increase the achievements of the depression classification system, audio, visual, and text descriptors are used. So, a deep learning-based depression classification model is developed to detect the person with depression from multimodal data. The EEG signal, Speech signal, video, and text are gathered from standard databases. Four stages of feature extraction take place. In the first stage, the features from the decomposed EEG signals are attained by the empirical mode decomposition (EMD) method, and features are extracted by means of linear and nonlinear feature extraction. In the second stage, the spectral features of the speech signals from the Mel-frequency cepstral coefficients (MFCC) are extracted. In the third stage, the facial texture features from the input video are extracted. In the fourth stage of feature extraction, the input text data are pre-processed, and from the pre-processed data, the textual features are extracted by using the Transformer Net. All four sets of features are optimally selected and combined with the optimal weights to get the weighted fused features using the enhanced mountaineering team-based optimization algorithm (EMTOA). The optimal weighted fused features are finally given to the hybrid attention-based dilated network (HADN). The HDAN is developed by combining temporal convolutional network (TCN) with bidirectional long short-term memory (Bi-LSTM). The parameters in the HDAN are optimized with the assistance of the developed EMTOA algorithm. At last, the classified output of depression is obtained from the HDAN. The efficiency of the developed deep learning HDAN is validated by comparing it with various traditional classification models.

Due to the absence of early facilities, a large population is dealing with stress, anxiety, and depression issues, which may have disastrous consequences, including suicide. Past studies revealed a direct relationship between the high engagement with social media and the increasing depression rate. This research initially creates a dataset with text, emoticons and image data, and then preprocessing is performed using diverse techniques. The proposed model in the research consists of three parts: first is textual bidirectional encoder representations from transformers (BERT), which is trained on only text data and also emoticons are converted into a textual form for easy processing; second is convolutional neural network (CNN), which is trained only on image data; and the third is the combination of best-performing models, i.e. hybrid of BERT and CNN (BERT-CNN), to work on both the text and images with enhanced accuracy. The results show the best accuracy with BERT, i.e. 97% for text data; for image data, CNN has attained the highest accuracy of 89%. Finally, the hybrid approach is compared with other combinations and previous studies; it achieved the best accuracy of 99% in the categorization of users into depressive and non-depressive based on multimodal data.

Depression is a widespread mental health issue with profound global impact, often leading to diminished life quality and increased suicide risk. Despite available treatments, many depression cases go unnoticed and untreated. This underscores the necessity for a precise, personalized model to predict depression severity and individual risk factors, utilizing machine learning on comprehensive, multimodal datasets. While previous efforts employing machine learning (ML) to gauge depression severity exist, their effectiveness has been curtailed by small datasets and a lack of personalization. To address this gap, we propose an advanced ML-based approach for predicting depression severity and identifying personalized risk factors. ML enhances the precision of depression severity assessments, facilitates personalized treatment strategies, and improves the identification of individual risk factors. In our study, we implemented, assessed, and compared five supervised ML algorithms-Linear Regression (LR), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Random Forest (RF), and Least Absolute Shrinkage and Selection Operator (LASSO)-known for their accuracy, interpretability, and computational efficiency. We utilized a multimodal dataset from the National Health and Nutrition Examination Survey (NHANES), encompassing demographic, dietary, socio-economic, lifestyle, medical, laboratory, and clinical data. The Random Forest algorithm proved to be the most effective, demonstrating an R-squared of 0.93, an explained variance score (EVS) of 0.93, a mean absolute error (MAE) of 0.51, a mean squared error (MSE) of 1.73, and a root mean squared error (RMSE) of 1.32. It effectively pinpointed both general and personalized risk factors for depression severity. Our model not only proves effective in predicting depression severity and identifying personalized risk factors but also shows promise for clinical application in assessment, diagnosis, treatment planning, and depression management.

Depression recognition (DR) using facial images, audio signals, or language text recordings has achieved remarkable performance. Recently, multimodal DR has shown improved performance over single-modal methods by leveraging information from a combination of these modalities. However, collecting high-quality data containing all modalities poses a challenge. In particular, these methods often encounter performance degradation when certain modalities are either missing or degraded. To tackle this issue, we present a generalizable multimodal framework for DR by aggregating feature disentanglement and privileged knowledge distillation. In detail, our approach aims to disentangle homogeneous and heterogeneous features within multimodal signals while suppressing noise, thereby adaptively aggregating the most informative components for high-quality DR. Subsequently, we leverage knowledge distillation to transfer privileged knowledge from complete modalities to the observed input with limited information, thereby significantly improving the tolerance and compatibility. These strategies form our novel Feature Disentanglement and Privileged knowledge Distillation Network for DR, dubbed Dis2DR. Experimental evaluations on AVEC 2013, AVEC 2014, AVEC 2017, and AVEC 2019 datasets demonstrate the effectiveness of our Dis2DR method. Remarkably, Dis2DR achieves superior performance even when only a single modality is available, surpassing existing state-of-the-art multimodal DR approaches AVA-DepressNet by up to 9.8% on the AVEC 2013 dataset.

No abstract available

Depression is a major psychological disorder with a growing impact worldwide. Traditional methods for detecting the risk of depression, predominantly reliant on psychiatric evaluations and self-assessment questionnaires, are often criticized for their inefficiency and lack of objectivity. Advancements in deep learning have paved the way for innovations in depression risk detection methods that fuse multimodal data. This paper introduces a novel framework, the Audio, Video, and Text Fusion-Three Branch Network (AVTF-TBN), designed to amalgamate auditory, visual, and textual cues for a comprehensive analysis of depression risk. Our approach encompasses three dedicated branches—Audio Branch, Video Branch, and Text Branch—each responsible for extracting salient features from the corresponding modality. These features are subsequently fused through a multimodal fusion (MMF) module, yielding a robust feature vector that feeds into a predictive modeling layer. To further our research, we devised an emotion elicitation paradigm based on two distinct tasks—reading and interviewing—implemented to gather a rich, sensor-based depression risk detection dataset. The sensory equipment, such as cameras, captures subtle facial expressions and vocal characteristics essential for our analysis. The research thoroughly investigates the data generated by varying emotional stimuli and evaluates the contribution of different tasks to emotion evocation. During the experiment, the AVTF-TBN model has the best performance when the data from the two tasks are simultaneously used for detection, where the F1 Score is 0.78, Precision is 0.76, and Recall is 0.81. Our experimental results confirm the validity of the paradigm and demonstrate the efficacy of the AVTF-TBN model in detecting depression risk, showcasing the crucial role of sensor-based data in mental health detection.

Motivated by depression's significant impact on global health, this work proposes MultiDepNet, a novel multi-modal interpretable depression detection system integrating visual, physiological, audio, and textual data. Through ded-icated feature extraction methods (MTCNN for video, TS-CAN for physiological, ResNet-18 for audio, and RoBERTa for text modalities) and a strategic fusion of modality-specific networks including CNN-RNN, Transformer, MLP, and ResNet-18, it achieves significant advancements in depression detection. Its performance, evaluated across four benchmark datasets (AVEC 2013, AVEC 2014, DAIC, and E-DAIC), demonstrates average MAE of 5.64, RMSE of 7.15, accuracy of 74.19%, precision of 0.7373, re-call of 0.7378, and F1 of 0.7376. It also implements a Multiviz-based interpretability mechanism that computes each modality's contribution to the model's performance. The results reveal the visual modality to be the most signifi-cant, contributing 37.88% towards depression detection.

Depression remains one of the most prevalent yet underdiagnosed mental health conditions worldwide. Traditional diagnostic tools often rely on subjective evaluations, which limit timely and scalable intervention. Recent advances in affective computing have enabled the use of multimodal datasuch as audio, visual, and text-for automated depression detection. This study proposes a cross-modal deep learning architecture that leverages attention-based fusion to integrate heterogeneous behavioral signals from limited DAIC-WOZ data. The model processes each modality through dedicated subnetworks and employs a multi-head cross-attention mechanism to learn inter-modality dependencies before final classification. Unlike prior approaches such as ACMA and FPTFormer, our model emphasizes lightweight deployment by reducing architectural complexity while maintaining competitive accuracy. Despite being trained on only 97 participants due to storage constraints, the system achieves $\mathbf{8 0 \%}$ accuracy and a macro-averaged $\mathbf{F}$ 1-score of $\mathbf{0. 7 8}$, demonstrating strong performance under data limitations. These findings highlight the feasibility of scalable, interpretable, and efficient AI frameworks for mental health screening in resourceconstrained environments.

Wearable and phone sensor data hold great potential for monitoring depression, yet effective integration of these diverse data sources remains challenging. Transforming these complex data into a learned embedding space provides a lower-dimensional representation that preserves essential temporal patterns while capturing the intricate inter-modal relationships. In this study, we evaluate how different fusion strategies for generating multimodal embeddings impact the effectiveness of clustering in identifying depression symptoms. We used a longitudinal dataset integrating physiological and social data such as electrocardiogram, accelerometer, respiration rate, and mobility/Bluetooth interaction data, collected over 35 days. An embedding-based approach using long short-term memory (LSTM) autoencoders was employed to learn latent space representations, followed by the application of K-Means and Gaussian Mixture Models (GMM) clustering algorithms to identify patterns within this learned space. Weekly Beck Depression Inventory-II (BDI-II) scores, held-out during training, served as the ground truth for performance evaluation. A custom metric, the BDI-Variance Ratio Clustering Score (BDI-VRCS), was developed to quantitatively assess clustering efficacy across different embedding spaces. Early fusion implementation with LSTM and GMM achieved the highest BDI-VRCS of 0.3309, outperforming both mid and late fusion strategies (0.112 and 0.132, respectively). This highlights the value of early integration of multimodal data, with social features playing a key role in capturing depressive symptoms.Clinical relevance— This study highlights the potential of integrating physiological and social data using multimodal fusion strategies to enhance depression monitoring and support the development of holistic, data-driven tools for early detection and personalized mental health interventions.

   A trending task of automatic psycho-emotional human state detection was studied in this work. A scientific interest to researches devoted to the automatic multimodal depression detection can arise out of the widespread of anxiety-depressive disorders and difficulties of their detection in primary health care. A specificity of the task was caused by its complexity, lack of data, imbalance of classes and inaccuracies in it. Comparative researches show that classification results on semi-automatic annotated data are higher than ones on automatic-annotated data. The proposed approach for depression detection combines a semi-automatic data annotation and deterministic machine learning methods with the utilization of several feature sets. To build our models, we utilized the multimodal Extended Distress Analysis Interview Corpus (E-DAIC) which consists of audio recordings, automatically extracted from these audio recordings texts and video feature sets extracted from video recordings as well as annotation including Patient Health Questionnaire (PHQ-8) scale for each recording. A semi-automatic annotation makes it possible to get the exact time stamps and speech texts to reduce the noisiness in the training data. In the proposed approach we use several feature sets, extracted from each modality (acoustic expert feature set eGeMAPS and neural acoustic feature set DenseNet, visual expert feature set OpenFace and text feature set Word2Vec). A complex processing of these features minimizes the effect of class imbalance in the data on classification results. Experimental researches with the use of mostly expert features (DenseNet, OpenFace, Word2Vec) and deterministic machine learning classification methods (Catboost) which have the property of interpretability of classification results yielded the experimental results on the E-DAIC corpus which are comparable with the existing ones in the field (68.0 % and 64.3 % for Weighted F1-measure (WF1) and Unweighted Average Recall (UAR) accordingly). The usage of a semi-automatic annotation approach and modalities fusion improved both quality of annotation and depression detection comparing to the unimodal approaches. More balanced classification results are achieved. The usage of deterministic machine learning classification methods based on decision trees allows us to provide an interpretability analysis of the classification results in the future due to their interpretability feature. Other methods of results interpretation like SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) also can be used for this purpose.

During psychiatric assessment, clinicians observe not only what patients report, but important nonverbal signs such as tone, speech rate, fluency, responsiveness, and body language. Weighing and integrating these different information sources is a challenging task and a good candidate for support by intelligence-driven tools—however this is yet to be realized in the clinic. Here, we argue that several important barriers to adoption can be addressed using Bayesian network modelling. To demonstrate this, we evaluate a model for depression and anxiety symptom prediction from voice and speech features in large-scale datasets (30,135 unique speakers). Alongside performance for conditions and symptoms (for depression, anxiety ROC-AUC = 0.842, 0.831 ECE = 0.018, 0.015; core individual symptom ROC-AUC > 0.74), we assess demographic fairness and investigate integration across and redundancy between different input modality types. Clinical usefulness metrics and acceptability to mental health service users are explored. When provided with sufficiently rich and large-scale multimodal data streams and specified to represent common mental conditions at the symptom rather than disorder level, such models are a principled approach for building robust assessment support tools: providing clinically-relevant outputs in a transparent and explainable format that is directly amenable to expert clinical supervision.

No abstract available

No abstract available

Depression, driven by growing societal pressures, significantly disrupts individuals’ physical and mental health. Automatic Depression Recognition (ADR) via facial videos has gained attention to enhance diagnostic accuracy and efficiency. However, extant methods often segment videos, losing long-term behavioral cues and introducing noise, while also exhibiting performance drops across diverse cultural and racial datasets. This study proposes a multimodal ADR approach encompassing three key components: (1) Long-term Depression Behavior Module (LDBM) employing a Transformer to capture extended depression cues, (2) Noisy Information Elimination (NIE) strategy leveraging LDBM attention scores to reduce noise and boost diagnostic precision, and (3) Multimodal Spatio-temporal Routing Feature Ensemble (MSRE) that fuses texture, Facial Action Primitives (FAPs), and Remote Photoplethysmography (rPPG) data for improved cross-dataset generalizability. Experiments on AVEC 2013, AVEC 2014, and a newly constructed CMDep dataset of 123 clinically diagnosed participants validate our method, achieving MAE/RMSE scores of 5.38/6.74, 5.09/6.83, and 5.59/8.03, respectively. The CMDep dataset includes facial expression and voice signals, with labels derived from BDI-II scores. Additionally, our method has been integrated into a user-friendly mobile application, providing a tool for real-time self-assessment of depression. This integration broadens the scope of depression detection, making it accessible to diverse populations worldwide.

Depression is a common mental health condition that impacts millions of people globally, its diagnosis often relies on subjective assessments. This project introduces a Multimodal Convolutional Neural Network (MCNN) framework tailored for depression detection, leveraging speech and eye movement data as input modalities. The project's foundation lies in combining advanced deep learning techniques with multimodal data analysis. Through the MCNN architecture, the system learns to extract discriminative features from speech recordings and eye movement patterns, enabling holistic assessments of individuals' mental health status. Data acquisition involves the collection and preprocessing of multimodal data samples from individuals both with and without depression. The MCNN model is trained on this curated dataset, exploiting the convolutional layers to process spatial features from eye movement images and temporal features from speech spectrograms concurrently. The data set is preprocessed using different algorithms of MCNN. We use MCFF’s algorithm, which represents the spectral envelope of the speech signal. Eyeball movement is processed by cluster-based algorithms that Convert raw eye movement data into a suitable format for analysis. This may involve segmenting the data into fixations and saccades, and representing them as sequences of gaze positions or heatmaps. The anticipated outcome of this project is a scalable, efficient, and accurate system for depression detection, capable of outperforming conventional diagnostic methods. Future directions include extending the MCNN framework to incorporate additional modalities, such as physiological data, and deploying the system in real-world clinical settings to facilitate early intervention and personalized treatment strategies.

Integrating physiological signals such as electroencephalogram (EEG), with other data such as interview audio, may offer valuable multimodal insights into psychological states or neurological disorders. Recent advancements with Large Language Models (LLMs) position them as prospective "health agents'' for mental health assessment. However, current research predominantly focus on single data modalities, presenting an opportunity to advance understanding through multimodal data. Our study aims to advance this approach by investigating multimodal data using LLMs for mental health assessment, specifically through zero-shot and few-shot prompting. Three datasets are adopted for depression and emotion classifications incorporating EEG, facial expressions, and audio (text). The results indicate that multimodal information confers substantial advantages over single modality approaches in mental health assessment. Notably, integrating EEG alongside commonly used LLM modalities such as audio and images demonstrates promising potential. Moreover, our findings reveal that 1-shot learning offers greater benefits compared to zero-shot learning methods.

Predicting the presence of major depressive disorder (MDD) using behavioural and cognitive signals is a highly non-trivial task. The heterogeneous clinical profile of MDD means that any given speech, facial expression and/or observed cognitive pattern may be associated with a unique combination of depressive symptoms. Conventional discriminative machine learning models potentially lack the complexity to robustly model this heterogeneity. Bayesian networks, however, may instead be well-suited to such a scenario. These networks are probabilistic graphical models that efficiently describe the joint probability distribution over a set of random variables by explicitly capturing their conditional dependencies. This framework provides further advantages over standard discriminative modelling by offering the possibility to incorporate expert opinion in the graphical structure of the models, generating explainable model predictions, informing about the uncertainty of predictions, and naturally handling missing data. In this study, we apply a Bayesian framework to capture the relationships between depression, depression symptoms, and features derived from speech, facial expression and cognitive game data collected at thymia.

No abstract available

Traditional depression detection methods typically rely on single-modal data, but these approaches are limited by individual differences, noise interference, and emotional fluctuations. To address the low accuracy in single-modal depression detection and the poor fusion of multimodal features from electroencephalogram (EEG) and speech signals, we have proposed a multimodal depression detection model based on EEG and speech signals, named the multi-head attention-GCN_ViT (MHA-GCN_ViT). This approach leverages deep learning techniques, including graph convolutional networks (GCN) and vision transformers (ViT), to effectively extract and fuse the frequency-domain features and spatiotemporal characteristics of EEG signals with the frequency-domain features of speech signals. First, a discrete wavelet transform (DWT) was used to extract wavelet features from 29 channels of EEG signals. These features serve as node attributes for the construction of a feature matrix, calculating the Pearson correlation coefficient between channels, from which an adjacency matrix is constructed to represent the brain network structure. This structure was then fed into a graph convolutional network (GCN) for deep feature learning. A multi-head attention mechanism was introduced to enhance the GCN's capability in representing brain networks. Using a short-time Fourier transform (STFT), we extracted 2D spectral features of EEG signals and mel spectrogram features of speech signals. Both were further processed using a vision transformer (ViT) to obtain deep features. Finally, the multiple features from EEG and speech spectrograms were fused at the decision level for depression classification. A five-fold cross-validation on the MODMA dataset demonstrated the model's accuracy, precision, recall, and F1 score of 89.03%, 90.16%, 89.04%, and 88.83%, respectively, indicating a significant improvement in the performance of multimodal depression detection. Furthermore, MHA-GCN_ViT demonstrated robust performance in depression detection and exhibited broad applicability, with potential for extension to multimodal detection tasks in other psychological and neurological disorders.

Many people around the world face mental health challenges, particularly students and professionals. Students often experience poor academic performance, difficulties concentrating, and memory issues, while professionals may face challenges with decision-making, communication, and confidence. Our system addresses these concerns by predicting mental health states like stress, anxiety, and depression through a standardized set of scored questions, with the cumulative score indicating the user's mental health status. The system also provides audio-based emotional state prediction using CNNs and Librosa for feature extraction and processing. Video-based emotional state prediction uses CNNs and OpenCV for video analysis. Additionally, the system includes a diary feature that tracks emotions throughout the day using BERT. At the end of each day, it summarizes the user's emotional state and maintains this data for five consecutive days. Afterward, a detailed report of emotional patterns is generated. The system also keeps a history of test predictions, enabling users to track their mental health trends over time. Personalized recommendations for psychiatrists are provided based on the results and user location. Comprehensive reports can be shared with family or friends for support and awareness, promoting early intervention and personalized mental health care.

Depression is a prevalent mental health disorder, and early detection is crucial for timely intervention. Traditional diagnostics often rely on subjective judgments, leading to variability and inefficiency. This study proposes a fusion model for automated depression detection, leveraging bimodal data from voice and text. Wav2Vec 2.0 and BERT pre-trained models were utilized for feature extraction, while a multi-scale convolutional layer and Bi-LSTM network were employed for feature fusion and classification. Adaptive pooling was used to integrate features, enabling simultaneous depression classification and PHQ-8 severity estimation within a unified system.Experiments on the CMDC and DAIC datasets demonstrate the model’s effectiveness. On CMDC, the F1 score improved by 0.0103 and 0.2017 compared to voice-only and text-only models, respectively, while RMSE decreased by 0.5186. On DAIC, the F1 score increased by 0.0645 and 0.2589, with RMSE reduced by 1.9901. These results highlight the proposed method’s ability to capture and integrate multi-level information across modalities, significantly improving the accuracy and reliability of automated depression detection and severity prediction.

Depression is a widespread mental health issue affecting diverse age groups, with notable prevalence among college students and the elderly. However, existing datasets and detection methods primarily focus on young adults, neglecting the broader age spectrum and individual differences that influence depression manifestation. Current approaches often establish a direct mapping between multimodal data and depression indicators, failing to capture the complexity and diversity of depression across individuals. This challenge includes two tracks based on age-specific subsets: Track 1 uses the MPDD-Elderly dataset for detecting depression in older adults, and Track 2 uses the MPDD-Young dataset for detecting depression in younger participants. The Multimodal Personality-aware Depression Detection (MPDD) Challenge aims to address this gap by incorporating multimodal data alongside individual difference factors. We provide a baseline model that fuses audio and video modalities with individual difference information to detect depression manifestations in diverse populations. This challenge aims to promote the development of more personalized and accurate de pression detection methods, advancing mental health research and fostering inclusive detection systems. More details are available on the official challenge website: https://hacilab.github.io/MPDDChallenge.github.io.

Depression is one of the most common mental illnesses, but few of the currently proposed in-depth models based on social media data take into account both temporal and spatial information in the data for the detection of depression. In this paper, we present an efficient, low-covariance multimodal integrated spatio-temporal converter framework called DepMSTAT, which aims to detect depression using acoustic and visual features in social media data. The framework consists of four modules: a data preprocessing module, a token generation module, a Spatial-Temporal Attentional Transformer (STAT) module, and a depression classifier module. To efficiently capture spatial and temporal correlations in multimodal social media depression data, a plug-and-play STAT module is proposed. The module is capable of extracting unimodal spatio-temporal features and fusing unimodal information, playing a key role in the analysis of acoustic and visual features in social media data. Through extensive experiments on a depression database (D-Vlog), the method in this paper shows high accuracy (71.53%) in depression detection, achieving a performance that exceeds most models. This work provides a scaffold for studies based on multimodal data that assists in the detection of depression.

Depression is a prevalent mental health disorder, and early detection is crucial for effective intervention. Recent advancements in eye-tracking technology and machine learning offer new opportunities for non-invasive diagnosis. This study aims to assess the performance of different machine learning algorithms in. predicting depression in a young sample using eye-tracking metrics. Eye-tracking data from 139 participants were recorded with an emotional induction paradigm in which each participant observed a set of positive and negative emotional stimuli. The data were analyzed to find differences between groups, where the most significant features were selected to train prediction models. The dataset was then split into training and testing sets using stratified sampling. Four algorithms support vector machines (SVM), random forest (RF), a multi-layer perceptron (MLP) neural network, and gradient boosting (GB) were trained with hyperparameter optimization and 5-fold cross-validation. The RF algorithm achieved the highest accuracy at 84%, followed by SVM, GB, and the MLP neural network. Performance metrics such as accuracy, recall, F1-score, precision recall area under the curve (PR-AUC), and Matthews Correlation Coefficient (MCC) were also used to evaluate the models. The findings suggest that eye-tracking metrics combined with machine learning algorithms can effectively identify depressive symptoms in the young, indicating their potential as non-invasive diagnostic tools in clinical settings.

Depression, a prevalent psychological disorder, increasingly affects individuals worldwide, leading to severe consequences such as a heightened risk of suicide when left undetected or untreated. Despite its widespread impact, depression often remains inadequately recognized within society, leaving many sufferers without necessary support. In this study, we leveraged machine learning techniques to address this issue, employing nine classifiers to detect depression using sociodemographic and psychosocial factors. Through three experiments utilizing different sampling techniques (SMOTE, SMOTETomek, and SMOTEETT), we identified key features using the SelectKBest method. Remarkably, the Light Gradient Boosting Machine (LightGBM) classifier, when applied with the SMOTEETT sampling technique, exhibited exceptional performance, achieving an accuracy of 0.9912 and outperforming other machine learning classifiers. Our findings underscore the effectiveness of machine learning approaches, particularly LightGBM, in detecting depression, highlighting their potential for early intervention and improved mental health outcomes.

Improving digital depression screening is important for combating the global mental health crisis. Textual data are promising for depression screening due to their many origins, but the variety presents screening challenges. To improve depression screening with textual data, we propose eXtreme Gradient Boosting (XGBoost) with Bayesian Optimization (BO). We experiment with three different objective functions to optimize our models. We apply our models to screen for depression with three disparate textual datasets containing features extracted from transcripts, SMS text messages, and typed replies. When compared to seven other machine learning methods, our XGBoost with BO models demonstrated impressive generalizability across the datasets, achieving average balanced accuracy scores of 0.60, 0.67, and 0.69 with transcripts, SMS text messages, and typed replies, respectively. Our feature importance assessment revealed that the most important features for these three text types were respectively negative emotion, youth, and love lexical category frequencies. Overall, our research presents a promising depression screening method that offers generalizability across text types, explainability, and computational efficiency.

Major depressive disorder (MDD) significantly impairs psychosocial functioning and reduces quality of life. Developing reliable, objective biomarkers for depression diagnosis remains a critical yet challenging task. Since Electroencephalography (EEG) allows for the examination of brain activity patterns associated with MDD, this work leverages deep learning techniques on emotion EEG data to bridge the gap between observable depression symptoms and underlying neural signatures for objective depression detection. EEG recordings were collected from 33 depressed patients (DPs) and 40 healthy controls (HCs) in response to happy, neutral, and sad emotional stimuli. We propose a hybrid Emotion EEG CNN-Transformer model (EMOCT) for DP-HC classification. By combining CNN and Transformer blocks, EMOCT effectively captures temporal, spectral, and spatial features, enabling a more comprehensive representation of brain activity related to depressed or healthy mental states. Our extensive experiments demonstrate that EMOCT outperforms other models in DP-HC classification, achieving accuracies of 85.97%, 82.83%, and 85.25% for happy, neutral, and sad emotion EEG data, respectively. The results highlight EMOCT’s potential as an effective and objective diagnostic tool for depression, paving the way for improved clinical assessment and management of the disorder.

Depression is a severe mental disorder that is associated with a persistent state of sadness, anhedonia, and psychosocial dysfunction. In its most severe form, it may culminate in suicidal ideation and behavior. Early diagnosis is crucial to attenuate adverse clinical outcomes. In this regard, recent procedures that include the analyses of speech acoustic parameters, facial features, and transcribed linguistic content have emerged as essential alternatives for identifying this disorder. The present study introduces a Multimodal Stacked Multilevel Deep Neural Network model applied to depression recognition in predisposed patients. This approach integrates base models into a metamodel, all of which are grounded in deep learning architectures for processing audiovisual and audio-textual information at multiple levels of representation. In addition, the solution incorporates specific strategies to ease the adverse effects associated with integrating various data sources such as class imbalance, high variance, and elevated dimensionality, thereby making the models more resilient to these challenges. In total, 11 metamodels were evaluated, and their results were highlighted, including those of the BiLSTM, DCNN, Transformer, and TCN architectures. In particular, the BiLSTM, which in the multimodal datasets D-Vlog and EDAIC achieved 75.0% and 60.2% Precision, 82.9% and 70.6% Recall, and 78.7% and 64.8% F1-score, respectively, attesting not only to the effectiveness of the proposed methodology, but also to its superiority over the main approaches reported in the literature, statistically confirmed in the D-Vlog dataset.

The prevalence of depression is escalating, especially among youth, which has become a critical mental health concern. Current assessment methods, relying heavily on questionnaires, clinical observations, and AI-driven analyses, are limited by their focus on single-event data, failing to encapsulate the nuanced expressions of depressive symptoms. Moreover, a significant oversight in existing research is the underutilization of electromyogram (EMG) alongside electroencephalogram (EEG) data, which could provide a more holistic view of unconscious body behaviors. Additionally, given the high variability of depression among individuals, traditional analysis models are in urgent need of refinement to accommodate the personality of different individuals. To address these limitations, we propose M3ADD, a novel benchmark for Automatic Depression Detection that employs a Multimodal, Multitask, and Multievent framework. We collected EEG and EMG data from 97 participants across varied events (interview, reading tasks, walking), coupled with standardized questionnaires assessing depression, wellbeing, and personality, enriching our multitask learning approach. Our benchmark recognition algorithm leverages multitask learning, channel and interactive attention mechanisms to synthesize event-specific and modal-specific features, enhancing adaptability to individual differences and improving data utilization efficiency. M3ADD surpasses existing models by achieving 87% accuracy in detecting depression and 95% accuracy in assessing wellbeing, providing a promising avenue for early identification.

Globally, 280 million are suffering from depressive disorders, one of the most common mental disorders referring to long periods of depressed mood that affect life negatively, often leading to suicide. Adolescents are especially vulnerable, with 35% of youth aged 12 to 17 in the United States having major or severe depression. The rate of depression in teenagers has doubled over the past decade, along with the youth suicide rate. Currently, self-report questionnaires such as the Patient Health Questionnaire are utilized to diagnose depression. Self-report questionnaires may be accurate for other population groups that are fully aware of the state of their mental health, and are willing to seek support. However, adolescents do not have the same ability to recognize their symptoms nor answer honestly in school-wide screening tests. In order to solve the aforementioned problem, a method that captures the true emotion without self-reports based on electroencephalogram (EEG) and galvanic skin response (GSR) is proposed. The novel system is composed of representation and transfer learning steps for the extraction of emotion-related features from EEG signals, and a 1D convolutional neural network for extraction in GSR signals. The extracted features are merged and outputted as points on the arousal-valence graph, where emotions can be detected. Through extensive experiments, the proposed model demonstrated exceptional performance. The best MAE was 18.4 when without GSR, 18.8 when without representation learning, and 17.2 when without the proposed equation, but improved to 16.4 in the proposed model.

No abstract available

Adolescent depression is a significant public health concern in low- and middle-income countries, including Kenya, where limited screening capacity contributes to under diagnosis. This study developed and validated a Bayesian Neural Network for predicting depression risk among adolescents, leveraging probabilistic inference to capture predictive uncertainty and complex nonlinear relationships. Secondary data for 2,192 adolescents aged 12-18 years were obtained from the Open Science Framework, incorporating psychosocial, demographic, and mental health measures. Significant predictors were identified using chi-square tests and point-biserial correlations, followed by forward feature selection. The proposed Bayesian Neural Network employed a three-hidden-layer feedforward architecture with ReLU and sigmoid activations, Gaussian priors on weights, and Bayes-by-Backprop Variational inference. Model performance was benchmarked against a Random Forest classifier using cross-validation. Results indicate that anxiety, loneliness, perceived social support, gratitude, positive youth development, academic self-perception, gender, academic form, financial status, parental education, and age are significant predictors of depression. The model achieved superior performance, with an accuracy of 80.13%, F1 score of 0.876, recall of 0.965, ROC-AUC of 0.813, and PR-AUC of 0.913, outperforming the Random Forest in most metrics except precision. Calibration analysis yielded a low Brier score (0.0693), indicating well-calibrated probabilistic predictions. These findings demonstrate the suitability of Bayesian Neural Networks for adolescent depression risk screening in resource-constrained settings, where uncertainty-aware decision support is critical.

These days, depression is a common illness worldwide which varies from usual mood fluctuations to challenges in everyday life. When depressive symptoms stay for long in the form of moderate or high intensity, it may result in becoming a serious health condition such as moving towards suicide. The advancement of machine learning can help technologies to build smart technologies to detect depression from different sources of data such as text. This work proposes a robust approach to detect depression from text messages using Long Short-Term Memory (LSTM)-based Neural Structured Learning (NSL). First of all, a text dataset is collected from the queries submitted in a Norwegian youth forum and then efficient features are obtained based on pre-defined handcrafted robust features developed by focusing on the symptoms of depression that were defined by medical practitioners and psychologists, rather than applying typical word frequency-based features (e.g., conventional term frequency-inverse document frequency (TF-IDF) and onehot) where the frequency of the words in the texts are focused but not the importance of the words. After that, LSTM-based NSL approach is applied as deep learning method to train the features discriminating depression and non-depression. The trained LSTM-based NSL is then utilized later to detect depression in the testing text messages. Besides, to explain the machine learning model's decision, a popular explainable Artificial Intelligence (XAI) algorithm is applied, which is Local Interpretable ModelAgnostic Explanations (LIME). The proposed approach shows the superiority against the traditional approaches on the dataset consists of Norwegian text, achieving the mean accuracy of 99%. Though it has been applied on Norwegian dataset, but the proposed concept can however be applied on other datasets as well, using the translated features.

Depression is a disease with severe consequences that affects millions of people, with the onset of the first symptoms being common in youth. It is essential to identify and treat individuals with depression as early as possible to prevent the losses caused by the disorder throughout life. However, the diagnostic criteria of depressive disorders for children/adolescents or adults is not differentiated, even though authors claim that the particularities of childhood must be considered. This may be why childhood depression is being underdiagnosed. Therefore, this work aims to discover the most significant features in diagnosing depression in children and adolescents through Machine Learning methods and the SHAP approach. Models with Machine Learning algorithms were developed, and the model with SVM presented the best results. The application of SHAP proved to be fundamental to deepen the understanding of this model. The experiments indicated that feelings of isolation, sadness, excessive worry, complaints about one’s appearance, resistance to academic tasks, and the mother’s schooling are the most significant features in predicting depression in children and adolescents. Such results can help to understand depression in these individuals and thus lead to appropriate treatment.

No abstract available

No abstract available

Untreated depression has been the major cause for suicide among youth. While psychotherapy, medication and other treatments have been successful in curing depression, majority of people suffering from depression do not actively seek, nor receive proper treatment. In many countries, mental health is not given as much importance as physical health and there is little or no awareness about mental disorders and their symptoms. Since depression has an effect on speech planning and production, we analyze a person's voice features to detect if they are depressed. We propose a model that processes the human voice signals given as input, converts it to a spectrogram image and builds a Neural Network Model that can be used to detect depression in humans. We implement this idea as a portable standalone device with a web interface as Proof of Concept.