Machine learning-based risk predictive models for depression in patients with diabetes: A systematic review and meta-analysis
By
Cai, Xingxin
Guo, Guiying
Zhou, Jun
Han, Mengqi
Cui, Yuanyuan
Chen, Zhenglin
March 30, 2026
Clinical Scorecard: Systematic Review and Meta-Analysis of Machine Learning Models for Predicting Depression Risk in Diabetic Patients
At a Glance
Category Detail
Condition Depression risk prediction in patients with diabetes mellitus (DM) Key Mechanisms Machine learning (ML) models including traditional regression, general ML, and deep learning to predict depression risk using clinical predictors Target Population Patients with diabetes mellitus Care Setting Clinical settings where depression risk assessment is relevant
Key Highlights
Pooled analysis of best-performing ML models showed good predictive performance with AUC of 0.822 (95% CI 0.789-0.858). Logistic regression was the most frequently used ML method for depression risk prediction in DM patients. Common predictors included age, sex, and body mass index (BMI), which are easily accessible in clinical practice. Guideline-Based Recommendations
Diagnosis
Utilize ML-based prediction models incorporating clinical predictors such as age, sex, and BMI to assess depression risk in diabetic patients. Management
Select ML models with demonstrated good predictive performance (AUC >0.8) for clinical decision support in depression risk management. Monitoring & Follow-up
Apply appropriate validation strategies (e.g., 5-fold or 10-fold cross-validation) to ensure model generalizability and avoid overfitting. Risks
Be aware of high risk of bias and heterogeneity among existing ML models; interpret predictions cautiously. Patient & Prescribing Data
Patients with diabetes mellitus at risk for depression
ML models can aid in early identification of depression risk to guide timely interventions, though clinical applicability requires careful validation.
Clinical Best Practices
Incorporate routinely collected clinical variables such as age, sex, and BMI in depression risk prediction models. Prefer ML models validated with robust cross-validation techniques to enhance reliability. Interpret ML model outputs in conjunction with clinical judgment due to variability and bias risks. References
