JMIR Publications

ABSTRACT

Background:

Depression in older adults is often underdiagnosed due to atypical symptom presentation and generational stigma, leading to delayed intervention. Early identification is critical to improve outcomes, but traditional approaches have limited predictive accuracy.

Objective:

This study aimed to develop and evaluate machine learning models to predict the two-year onset of depression in older adults using data from the English Longitudinal Study of Ageing (ELSA), and to identify key risk factors that contribute to depression onset.

Methods:

Data were drawn from four consecutive waves of ELSA, including participants aged 50 years and above who were free of baseline depression and prior psychiatric diagnoses. Over 120 sociodemographic, psychological, and health-related features were analyzed. We applied a range of traditional classifiers (Random Forest, XGBoost, Support Vector Machines) and deep tabular models (TabTransformer, TabNet). Model performance was assessed using area under the receiver operating characteristic curve (AUROC) and F1-score. Model interpretability was examined using SHapley Additive exPlanations (SHAP) to identify the most influential features.

Results:

Across waves, the best-performing models achieved mean AUROC scores of 0.72–0.73, with a peak of 0.75 in the highest-performing wave. SHAP consistently identified age, loneliness, sleep disturbances, and low social engagement as strong predictors of depression onset. Traditional machine learning models (Random Forest, XGBoost, Support Vector Machines) generally outperformed deep learning models for this task.

Conclusions:

Our findings demonstrate the feasibility of using machine learning to predict future depression onset in older adults with moderate accuracy. The identification of consistent risk factors highlights opportunities for developing targeted clinical screening tools and preventive interventions. This study provides novel evidence on depression prediction in the UK context, leveraging longitudinal data from ELSA, and contributes to advancing digital mental health research for aging populations.