Related Experiment Video
Updated: Jan 15, 2026
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Prediction model for depression risk in middle-aged and elderly patients with metabolic syndrome: a nomogram and
Jiahao Chen1, Yisi Lin2, Rui Hu3
1Department of Neurology, Affiliated Jinhua Hospital, Zhejiang University School of Medicine, Jinhua, China.
Background:
Individuals with metabolic syndrome (MetS) are more prone to depression, which is a significant complication impacting quality of life. This research seeks to create and validate predictive models for assessing depression risk in patients with MetS.
Methods:
Data from the 2011 (baseline) and 2015 waves of the China Health and Retirement Longitudinal Study (CHARLS) were employed in this study. By excluding variables with more than 20% missing values, 38 features, such as demographic information, lifestyle factors, comorbidities, health status indicators, and blood test information, were included. The Least Absolute Shrinkage and Selection Operator (LASSO) identified 11 key contributors, and 6 machine learning (ML) models were employed to determine the best depression risk in patients with MetS. Furthermore, the 2015 CHARLS data were included as a temporal validation cohort.
Results:
In the 2011 CHARLS data, 5204 patients with MetS were analyzed, of whom 2543 (48.6%) exhibiting depression as indicated by a CESD-10 score of 10 or higher. 11 factors were selected to develop six ML models. The logistic regression (LR) (AUC: 0.749, 95% CI: 0.725-0.773) and Extreme Gradient Boosting (XGBoost) (AUC: 0.749, 95% CI: 0.725-0.773) models showed the same predictive ability in the test set. Utilizing grid search optimization, the XGBoost model attained an AUC of 0.737 (95% CI: 0.714-0.760) on the validation set.
Conclusion:
The nomogram and SHAP visualization provide reliable tools for predicting depression in patients with MetS. The clinical utility of models applying LR and XGBoost is noteworthy, offering crucial insights for earlier detection and preventative actions for community staff and doctors.
