Predictive Model for Depression Risk in Middle-Aged and Older Adults with Metabolic Syndrome

Overview

This study developed a machine learning-based predictive model for depression risk among middle-aged and older adults with metabolic syndrome (MetS) using data from the China Health and Retirement Longitudinal Study (CHARLS). The model identified key risk factors and demonstrated robust predictive accuracy, validated with a temporal cohort.

Background

Depression is a prevalent mental health disorder with significant global impact, particularly among individuals with metabolic syndrome, who exhibit higher depression risk and mortality. Metabolic syndrome encompasses metabolic abnormalities that increase cardiovascular and diabetes risks. Prior studies have linked factors such as age, gender, education, sleep, and chronic pain to depression in metabolic disorders, but the mechanisms remain unclear. Machine learning offers advanced capabilities to analyze complex datasets and improve depression risk prediction in this population.

Data Highlights

The study included 5,205 participants with MetS from CHARLS wave 1 for model development and 1,943 participants from wave 3 for validation. MetS was defined by ATP III criteria requiring at least three of five metabolic abnormalities. Depression was assessed using the CESD-10 scale, with scores ≥10 indicating clinically significant depression. Predictor variables were selected from five domains based on literature review.

Key Findings

• The machine learning model effectively identified individuals with MetS at high risk for depression using demographic, clinical, and lifestyle variables.
• Key predictors included older age, female sex, low educational attainment, sleep duration, and presence of chronic pain.
• The model demonstrated strong generalizability and predictive accuracy when validated on a temporal cohort from CHARLS wave 3.
• Integration of interpretable machine learning techniques facilitated understanding of variable importance and risk stratification.
• The study highlighted the potential role of glymphatic system impairment linking metabolic dysfunction and depression.

Clinical Implications

Clinicians should consider comprehensive risk assessment for depression in patients with metabolic syndrome, incorporating demographic and lifestyle factors identified by the model. Early identification through such predictive tools can enable timely interventions and personalized management to reduce depression burden and associated mortality in this high-risk group.

Conclusion

This study successfully developed and validated a machine learning-based predictive model for depression risk among middle-aged and older adults with metabolic syndrome, providing a valuable tool for early detection and targeted intervention in clinical practice.

References

1. Global Burden of Disease Study 2019 -- Depression Impact
2. ATP III Criteria -- Metabolic Syndrome Definition
3. CHARLS Study Protocol -- China Health and Retirement Longitudinal Study
4. CESD-10 Validation -- Depression Assessment in Chinese Adults
5. Hong et al. 2020 -- ML Depression Risk Prediction in Elderly
6. Zheng et al. 2021 -- ML Model for Depression in Chronic Disease