Scholarlink

Background <p>Adolescents living with HIV (ALWHIV) face a high prevalence of depression. In Uganda, limited access to mental health services exacerbates depression, necessitating innovative solutions for early detection and intervention. Machine learning (ML) provides a promising approach for identifying adolescents at risk of depression to facilitate early intervention. This study aims to use machine learning to predict depression among ALWHIV in Uganda.</p> Methods <p>This study analyzed cross-sectional data from 833 ALWHIV aged 10–17 years in Southern Uganda. Depression was assessed using the Children’s Depression Inventory (CDI) –short version. Seven ML models, including Random Forest, Logistic Regression, Gradient Boosting, decision tree, XGBoost, Least Absolute Shrinkage and Selection Operator (LASSO), and Support Vector Machine, were evaluated using stratified 10-fold cross-validation. Model performance metrics included Area under receiver operating characteristic curve (AUROC), Area under precision-recall curve (AUPRC), accuracy, sensitivity, specificity, and F1-score. SHapley Additive exPlanations (SHAP) analysis was used to interpret feature importance.</p> Results <p>The median age was 13 years, and the majority of the participants were females (55.5%). The prevalence of depression was 30.97%. The Random Forest model achieved the highest performance (AUROC = 0.79, AUPRC = 0.66). Key predictors of depression included hopelessness, HIV stigma, shame, self-esteem, teacher support, and caregiver support. SHAP analysis highlighted psychosocial challenges as primary predictors of depression, emphasizing the need for psychosocial-based interventions.</p> Conclusion <p>Our study results demonstrated good performance of ML in predicting depression among ALWHIV. However, we acknowledge that the absence of external validation limits the generalizability of these findings. Future research should seek to validate these models in independent cohorts of ALWHIV to confirm their robustness and applicability across different settings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Using Machine Learning to Predict Depression among Adolescents Living with HIV in Uganda

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月21日

查看全文
  • Vicent Ssentumbwe,
  • Francis Matovu,
  • Phionah Namatovu,
  • Claire Najjuuko,
  • Samuel Kizito,
  • Josephine Nabayinda,
  • Portia Nartey,
  • Flavia Namuwonge,
  • Proscovia Nabunya,
  • Chenyang Lu,
  • Massy Mutumba,
  • Fred. M Ssewamala

摘要

Background

Adolescents living with HIV (ALWHIV) face a high prevalence of depression. In Uganda, limited access to mental health services exacerbates depression, necessitating innovative solutions for early detection and intervention. Machine learning (ML) provides a promising approach for identifying adolescents at risk of depression to facilitate early intervention. This study aims to use machine learning to predict depression among ALWHIV in Uganda.

Methods

This study analyzed cross-sectional data from 833 ALWHIV aged 10–17 years in Southern Uganda. Depression was assessed using the Children’s Depression Inventory (CDI) –short version. Seven ML models, including Random Forest, Logistic Regression, Gradient Boosting, decision tree, XGBoost, Least Absolute Shrinkage and Selection Operator (LASSO), and Support Vector Machine, were evaluated using stratified 10-fold cross-validation. Model performance metrics included Area under receiver operating characteristic curve (AUROC), Area under precision-recall curve (AUPRC), accuracy, sensitivity, specificity, and F1-score. SHapley Additive exPlanations (SHAP) analysis was used to interpret feature importance.

Results

The median age was 13 years, and the majority of the participants were females (55.5%). The prevalence of depression was 30.97%. The Random Forest model achieved the highest performance (AUROC = 0.79, AUPRC = 0.66). Key predictors of depression included hopelessness, HIV stigma, shame, self-esteem, teacher support, and caregiver support. SHAP analysis highlighted psychosocial challenges as primary predictors of depression, emphasizing the need for psychosocial-based interventions.

Conclusion

Our study results demonstrated good performance of ML in predicting depression among ALWHIV. However, we acknowledge that the absence of external validation limits the generalizability of these findings. Future research should seek to validate these models in independent cohorts of ALWHIV to confirm their robustness and applicability across different settings.