Related Experiment Video
Updated: Aug 2, 2025
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.6K
Interpretable Machine Learning for Fall Prediction Among Older Adults in China
Xiaodong Chen1, Lingxiao He1, Kewei Shi1
1Center for Aging and Health Research, School of Public Health, Xiamen University, Xiamen, China.
American Journal of Preventive Medicine
|April 22, 2023
Summary
Accurate fall risk prediction models are crucial for community-dwelling older adults in China. This study developed models using machine learning, identifying fall history as a key predictor for falls and fall-related injuries.
Area of Science:
- Gerontology
- Public Health
- Artificial Intelligence in Healthcare
Background:
- Falls in older adults pose significant health risks.
- Existing fall risk prediction models lack accuracy for community-dwelling older Chinese adults.
Purpose of the Study:
- To develop accurate prediction models for falls and fall-related injuries.
- To identify key risk factors for falls in this population.
Main Methods:
- Utilized data from 5,818 participants in the China Health and Retirement Longitudinal Study (2015 and 2018).
- Employed five machine learning algorithms to build 3-year risk prediction models.
- Applied SHapley Additive exPlanations for model interpretability.
Main Results:
- Logistic regression models demonstrated the best performance (AUC 0.739 for falls, 0.757 for injuries).
- Previous fall experience was the most significant predictor for both outcomes.
- Key factors included activities of daily living, depressive symptoms, grip strength, and sleep duration.
Conclusions:
- Developed models show promise for identifying high-risk older adults for targeted interventions.
- Fall prevention strategies should address fall history, physical function, psychological state, and home environment.
