JMIR Publications

ABSTRACT

Background:

The prevalence of adolescent mental health conditions, like depression and anxiety, has significantly increased. Despite the potential of machine learning (ML), there is a shortage of models that utilize real-world data (RWD) to enhance early detection and intervention for these conditions.

Objective:

To identify depression and anxiety in adolescents through RWD and social determinants of health (SDoH).

Methods:

We analyzed RWD of adolescents aged 10-17 years. We considered various factors such as demographics, prior diagnoses, prescribed medications, medical procedures, and laboratory measurements recorded before the onset of anxiety or depression. We connected the clinical data 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 it using the nested cross-validation technique. To interpret the model predictions, we utilized the SHapley Additive exPlanation (SHAP) method.

Results:

Our cohort included 52,054 adolescents, identifying 12,572 with anxiety, 7,812 with depression, and 14,019 with either condition. The models achieved AUC values of 0.80 for anxiety, 0.81 for depression, and 0.78 for both combined. Excluding SDoH data had minimal impact on model performance. SHAP analysis identified gender, race, educational attainment, and various medical factors as key predictors for anxiety and depression.

Conclusions:

This study highlights the potential of ML in identifying depression and anxiety in adolescents early on using RWD. By utilizing RWD, healthcare providers could more precisely identify at-risk adolescents and intervene earlier. This may lead to improved mental health outcomes.