Related Experiment Video
Updated: Aug 7, 2025
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.6K
Development and Assessment of Assisted Diagnosis Models Using Machine Learning for Identifying Elderly Patients With
Xue Wang1, Fengchun Yang2, Mingwei Zhu3,4,5
1Department of Clinical Nutrition, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China.
Journal of Medical Internet Research
|March 14, 2023
Summary
Machine learning models can now help identify malnutrition in older adults. Key indicators like BMI and weight loss are crucial for early detection and personalized treatment plans.
Area of Science:
- Gerontology
- Medical Informatics
- Nutritional Science
Background:
- Older adults face increased malnutrition risks, impacting clinical outcomes.
- Early identification of malnutrition is critical for effective patient management.
Purpose of the Study:
- To develop machine learning (ML) models for assisted diagnosis of malnutrition in older patients.
- To identify key predictors for individualized malnutrition treatment.
Main Methods:
- Reanalyzed a multicenter cohort of 2660 older patients.
- Applied 5 ML algorithms to identify malnutrition based on Global Leadership Initiative on Malnutrition (GLIM) criteria.
- Utilized Shapley additive explanations to explore feature importance.
Main Results:
- ML models accurately identified malnutrition in older patients.
- Top models (Light Gradient Boosting Machine, Extreme Gradient Boosting, Random Forest) achieved over 91.5% accuracy in external validation.
- Body Mass Index (BMI), weight loss, and calf circumference were identified as strong predictors.
Conclusions:
- Developed effective ML models for malnutrition diagnosis in older adults using GLIM criteria.
- Shapley additive explanations provide valuable reference cutoff values for malnutrition identification.
