Scholarlink

Background <p>With the increasingly serious trend of population aging in China, falls have become a critical public health issue endangering the health of older adults. The objective of this study was to develop an explainable machine learning model to elucidate the key factors influencing falls among elderly patients with chronic diseases.</p> Methods <p>This study was based on the 2018 follow-up data of the Chinese Longitudinal Health Longevity Survey (CLHLS), selecting aged 65 years and older patients with chronic diseases as the research subjects. This study established ten machine learning models for predicting falls in elderly patients with chronic diseases, including the Logistic Regression model, Random Forest (RF) model, K-Nearest Neighbor (KNN) model, Support Vector Machine (SVM) model, Gradient Boosting Machine (GBM) model, Neural Network (NNET) model, Extreme Gradient Boosting (Xgboost) model, Light Gradient Boosting Machine (LightGBM) model, Category Feature Gradient Boosting (CatBoost) model, and Adaptive Boosting (Adaboost) model. The receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were utilized to assess the model, while the importance of optimal model features was analyzed using the SHapley Additive exPlanations (SHAP) algorithm to enhance model transparency and explanation.</p> Results <p>Ten characteristic variables were determined by Lasso regression analysis and multivariable logistic regression to build the machine learning model. Model comparisons showed that GBM performed best, with an AUC of 0.844, accuracy of 0.794, and specificity of 0.967. SHAP analysis revealed that the top three characteristics of contribution included self-rated health status, basic activities of daily living (BADL) disorder and housing damage.</p> Conclusions <p>Among the machine learning models established based on the CLHLS database that can predict falls in elderly patients with chronic diseases, the GBM model demonstrates superior overall performance. The SHAP algorithm enhances the interpretability of GBM model.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on fall prediction in elderly patients with chronic diseases based on explainable machine learning: an aging perspective

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月9日

查看全文
  • Qin Zhang,
  • Yuting Yang,
  • Qiyan Hou,
  • Qingying Shi,
  • Yaolin Yi,
  • Xinyan Gan,
  • Xiang Gao

摘要

Background

With the increasingly serious trend of population aging in China, falls have become a critical public health issue endangering the health of older adults. The objective of this study was to develop an explainable machine learning model to elucidate the key factors influencing falls among elderly patients with chronic diseases.

Methods

This study was based on the 2018 follow-up data of the Chinese Longitudinal Health Longevity Survey (CLHLS), selecting aged 65 years and older patients with chronic diseases as the research subjects. This study established ten machine learning models for predicting falls in elderly patients with chronic diseases, including the Logistic Regression model, Random Forest (RF) model, K-Nearest Neighbor (KNN) model, Support Vector Machine (SVM) model, Gradient Boosting Machine (GBM) model, Neural Network (NNET) model, Extreme Gradient Boosting (Xgboost) model, Light Gradient Boosting Machine (LightGBM) model, Category Feature Gradient Boosting (CatBoost) model, and Adaptive Boosting (Adaboost) model. The receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were utilized to assess the model, while the importance of optimal model features was analyzed using the SHapley Additive exPlanations (SHAP) algorithm to enhance model transparency and explanation.

Results

Ten characteristic variables were determined by Lasso regression analysis and multivariable logistic regression to build the machine learning model. Model comparisons showed that GBM performed best, with an AUC of 0.844, accuracy of 0.794, and specificity of 0.967. SHAP analysis revealed that the top three characteristics of contribution included self-rated health status, basic activities of daily living (BADL) disorder and housing damage.

Conclusions

Among the machine learning models established based on the CLHLS database that can predict falls in elderly patients with chronic diseases, the GBM model demonstrates superior overall performance. The SHAP algorithm enhances the interpretability of GBM model.