Prediction model for depressive symptoms among overweight and obese adults in Qinghai Province
10.19485/j.cnki.issn2096-5087.2026.09.007
- Author:
LI Tong
;
HAN Shukui
;
REN Yitao
;
FU Lulu
;
FAN Pengcheng
;
CHEN Hongru
- Publication Type:Journal Article
- Keywords:
overweight;
obesity;
depressive symptoms;
random forest;
machine-learning algorithms;
prediction model
- From:
Journal of Preventive Medicine
2026;38(9):895-899
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct a prediction model for depressive symptoms among overweight and obese adults based on machine-learning algorithms.
Methods:Residents aged ≥18 years with body mass index ≥24.0 kg/m2 were recruited from three autonomous counties (districts) in Qinghai Province using the convenience sampling method from July 2024 to August 2025. Data on age, gender and the number of common chronic diseases were collected through questionnaire surveys. The Pittsburgh Sleep Quality Index, International Physical Activity Questionnaire and Patient Health Questionnaire-9 were used to assess sleep disorders, physical activity level and depressive symptoms, respectively. The participants were randomly divided into a training set and a validation set at a ratio of 7∶3. LASSO regression, Boruta algorithm and random forest model were applied to screen predictive factors. Five prediction models including random forest, logistic regression, decision tree, extreme gradient boosting and support vector machine were established. The prediction performance of the models was compared using the area under the receiver operating characteristic curve (AUC), Brier score, and decision curve analysis. SHAP values were adopted to evaluate the contribution of each variable to the model.
Results:A total of 1 065 overweight and obese adults were investigated, with a median age of 55.00 (interquartile range, 15.00) years. There were 498 males (46.76%) and 567 females (53.24%). Depressive symptoms were detected in 448 overweight and obese adults, with a detection rate of 42.07%. Annual household income, ethnicity, number of common chronic diseases, sleep disorders and physical activity level were identified as predictive factors for depressive symptoms among overweight and obese adults. Among the five established prediction models, the random forest model showed relatively optimal predictive performance. In the training set, the AUC value, accuracy, sensitivity, specificity, F1-score and Brier score were 0.839, 0.769, 0.682, 0.833, 0.713 and 0.168, respectively; the corresponding values in the validation set were 0.725, 0.646, 0.560, 0.708, 0.570 and 0.208, respectively. The maximum clinical net benefit of 0.277 was obtained at a threshold probability of 25%. Variable importance ranking indicated that sleep disorders contributed the most to the random forest model.
Conclusion:The random forest model constructed in this study has good discrimination and clinical utility for predicting the risk of depressive symptoms among overweight and obese adults.
- Full text:2026111215172807684青海省超重肥胖成年人抑郁症状预测模型研究.pdf
