Machine learning identifies depression risk in older adults with chronic diseases: Clarifying shared risk factors stratified by cognitive impairment status

BACKGROUND: The prevalence of depression is higher among older adults with chronic diseases and cognitive impairment than the general population. The comorbidity of cognitive impairment and chronic diseases significantly impacts the lives of these patients. This study aims to develop machine learning models to identify depression risk among older adults with chronic illnesses across different levels of cognitive impairment. METHODS: Data were derived from the Chinese Longitudinal Healthy Longevity Survey (n = 5798). The XGBoost algorithm was used to train and construct models based on data from 5798 participants. Model interpretability was enhanced using SHapley Additive exPlanations. To improve practical applicability, each model was further simplified based on feature importance. RESULTS: The accuracy of the three models ranged from 0.755 to 0.767, and all Brier scores were 0.160 or lower, indicating good predictive performance. 'Feeling energetic' was the most important predictor in both the cognitively unimpaired and mildly impaired groups, whereas 'sleep duration per day' was the top predictor in the severely impaired group. CONCLUSION: The proposed models demonstrated promising performance in identifying depression risk among older adults with chronic illnesses across different levels of cognitive impairment. The findings suggest that key predictors of depression may differ according to cognitive status, highlighting the importance of cognition-stratified screening strategies. Further external validation is needed before implementation in routine clinical practice.