JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Explainable machine learning model for pre-frailty risk assessment in community-dwelling older adults

  • 0School of Biomedical Engineering Tsinghua University Beijing China.
Health Care Science +

|

December 30, 2024

|

December 30, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

This study developed an explainable machine learning model to assess pre-frailty risk in older adults. The model accurately identifies individuals at risk, utilizing factors like living city, BMI, and peak expiratory flow.

Area Of Science

  • Gerontology
  • Artificial Intelligence
  • Public Health

Background

  • Frailty in older adults is a significant concern, associated with increased health risks and reduced quality of life.
  • Pre-frailty, a precursor to frailty, is identifiable and potentially reversible, yet its assessment and determinants remain challenging.
  • Community-dwelling older adults are a key population for pre-frailty interventions.

Purpose Of The Study

  • To develop and validate an explainable machine learning model for assessing pre-frailty risk.
  • To identify key determinants contributing to pre-frailty risk in older adults.
  • To provide a comprehensive framework for understanding and addressing pre-frailty.

Main Methods

  • Utilized data from 3141 adults aged 60+ from the China Health and Retirement Longitudinal Study.
  • Defined pre-frailty using one or two criteria from the physical frailty phenotype scale.
  • Employed recursive feature elimination and a stacking-CatBoost distillation module for model construction and validation on a 20% holdout set.

Main Results

  • Developed a distilled CatBoost model using 57 predictive features from 2508 older adults.
  • Achieved a high discrimination accuracy (AUROC: 0.7560) on the holdout data.
  • Identified living city, Body Mass Index (BMI), and peak expiratory flow (PEF) as the top three risk contributors.

Conclusions

  • Developed an accurate and interpretable framework for pre-frailty risk assessment using advanced machine learning.
  • The model integrates a wide array of features for a nuanced understanding of pre-frailty.
  • This framework supports comprehensive pre-frailty risk evaluation in community-dwelling older adults.

Keywords: