Machine Learning-Based Model for Grip Strength Prediction in Healthy Adults: A Nationwide Dataset-Based Study

Area of Science:

• Health Sciences
• Biomedical Engineering
• Data Science

Background:

• Handgrip strength (HGS) is a key indicator of overall health and functional status.
• Accurate estimation of HGS is crucial for early diagnosis and personalized interventions.
• Existing methods for HGS assessment may have limitations in large-scale population studies.

Purpose of the Study:

• To develop and compare machine learning models for estimating HGS in healthy adults.
• To identify the most accurate and generalizable model for HGS prediction.
• To explore key predictors influencing HGS estimation.

Main Methods:

• Analysis of data from 21,147 Korean adults (aged >19 years) from 2014-2019.
• Development and evaluation of Polynomial Regression (PR), Multilayer Perceptron (MLP), and Extreme Gradient Boosting (XGBoost) models.
• Utilized demographic, anthropometric, and physical activity data as predictors; assessed models using RMSE and R².

Main Results:

• HGS varied significantly by gender, age, and hand dominance, with males and younger individuals exhibiting higher strength.
• The XGBoost model achieved the highest predictive accuracy and generalizability, with an R² of 0.717.
• Key predictors for HGS in the XGBoost model included weight, age, height, and waist circumference.

Conclusions:

• The XGBoost model demonstrated superior performance over PR and MLP for HGS estimation.
• Accurate HGS estimation using machine learning holds significant potential for clinical applications.
• This approach can support early disease detection, targeted interventions, and personalized health goal setting.