Prediction models of grip strength in adults above 65 years using Korean National Physical Fitness Award Data from 2009 to 2019.

Purpose: We aimed to determine the best machine learning (ML) regression model for predicting grip strength in adults above 65 years using various independent variables, such as body composition, blood pressure, and physical performance.

Methods: The data comprised 107,290 participants, of whom 33.3% were male and 66.7% were female in Korean National Fitness Award Data from 2009 to 2019. The dependent variable was grip strength, which was calculated as the mean of right and left grip strength values.

Results: The results showed that the CatBoost Regressor had the lowest mean squared error (M [Formula: see text] SE:16.659 ± 0.549) and highest R value (M [Formula: see text] SE:0.719 ± 0.009) among the seven prediction models tested. The importance of independent variables in facilitating model learning was also determined, with the Figure-of-8 walk test being the most significant. These findings suggest that walking ability and grip strength are closely related, and the Figure-of- 8 walk test is a reasonable indicator of grip strength in older adults.

Conclusion: The findings of this study can be used to develop more accurate predictive models of grip strength in older adults.