Frontiers in psychiatry · 2026
Machine learning-based identification of anxiety symptoms in Chinese community-dwelling older adults: a comparative study of six algorithms with SHAP analysis and nomogram development.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Li Peiyue, Yan Chunxiao, Li Jianbo, Zheng Qimi, Zhang Huiqi
Аннотация
BACKGROUND: Anxiety disorders are common among older adults but remain underrecognized in community settings, particularly in China where mental health resources are scarce. This study aimed to develop and compare multiple machine learning models for identifying anxiety symptoms in Chinese community-dwelling older adults and to construct a practical clinical tool.
METHODS: A cross-sectional study included 5,331 community-dwelling older adults (aged ≥65 years) from Shenzhen, China. Anxiety symptoms were assessed using the Generalized Anxiety Disorder-7 (GAD-7) scale, with a score ≥5 indicating clinically significant anxiety. Six machine learning algorithms-Lasso, Random Forest, XGBoost, Decision Tree, Support Vector Machine, and Gradient Boosting Machine (GBM)-were trained and compared. Model performance was evaluated using area under the curve (AUC), sensitivity, specificity, and accuracy. SHapley Additive exPlanations (SHAP) was employed for model interpretability, and a nomogram was developed for clinical application.
RESULTS: The prevalence of anxiety symptoms was 11.3% (605/5,331). GBM achieved the highest AUC of 0.900 (sensitivity = 0.890, specificity = 0.762), comparable to XGBoost (AUC = 0.897) and Lasso (AUC = 0.893) with no significant differences (all P > 0.05). Feature combination analysis revealed that the "Psychological + Clinical" set achieved optimal performance (AUC = 0.903, PR-AUC = 0.617). SHAP analysis identified depressive symptoms (PHQ-9), insomnia (ISI), and loneliness (ULS-6) as the top three risk indicators. A nomogram incorporating nine predictors demonstrated good clinical utility.
CONCLUSIONS: Machine learning models, particularly GBM, showed excellent performance in identifying anxiety symptoms among Chinese community-dwelling older adults. The model demonstrates promising internal validity but remains internally validated only; clinical implementation is premature without external validation in independent cohorts.
METHODS: A cross-sectional study included 5,331 community-dwelling older adults (aged ≥65 years) from Shenzhen, China. Anxiety symptoms were assessed using the Generalized Anxiety Disorder-7 (GAD-7) scale, with a score ≥5 indicating clinically significant anxiety. Six machine learning algorithms-Lasso, Random Forest, XGBoost, Decision Tree, Support Vector Machine, and Gradient Boosting Machine (GBM)-were trained and compared. Model performance was evaluated using area under the curve (AUC), sensitivity, specificity, and accuracy. SHapley Additive exPlanations (SHAP) was employed for model interpretability, and a nomogram was developed for clinical application.
RESULTS: The prevalence of anxiety symptoms was 11.3% (605/5,331). GBM achieved the highest AUC of 0.900 (sensitivity = 0.890, specificity = 0.762), comparable to XGBoost (AUC = 0.897) and Lasso (AUC = 0.893) with no significant differences (all P > 0.05). Feature combination analysis revealed that the "Psychological + Clinical" set achieved optimal performance (AUC = 0.903, PR-AUC = 0.617). SHAP analysis identified depressive symptoms (PHQ-9), insomnia (ISI), and loneliness (ULS-6) as the top three risk indicators. A nomogram incorporating nine predictors demonstrated good clinical utility.
CONCLUSIONS: Machine learning models, particularly GBM, showed excellent performance in identifying anxiety symptoms among Chinese community-dwelling older adults. The model demonstrates promising internal validity but remains internally validated only; clinical implementation is premature without external validation in independent cohorts.