Previously submitted to: Journal of Medical Internet Research (no longer under consideration since Aug 17, 2023)
Date Submitted: May 9, 2023
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Automatic Personalized Physical Activity Prediction Model Based on the National Health Insurance Service Health Examination Data: A Machine Learning Approach
ABSTRACT
Background:
There has been growing public interest in maintaining personal health and preventing diseases through the adoption of healthy lifestyles. Within this paradigm, physical activity (PA) prescriptions are critical in facilitating healthy habits after undergoing health examinations.
Objective:
This study aimed to construct PA prediction models using clinical biomarkers that are commonly measured in clinical practice.
Methods:
We used the data from patients aged ≥20 years who underwent routine health screening in Korea from January 2021 to December 2021. Prediction models based on several machine learning algorithms, such as random forest classification, bagged decision trees, and gradient boosting (GB), were established.
Results:
A total of 30,769 patients who met the inclusion criteria were analyzed in the study. Of them, 13,653 were men and 17,116 were women. The prediction models using the GB algorithm exhibited the best performance, in particular, the GB-based models with female data only, with accuracy values close to or greater than 0.8. For these prediction models, the area under the receiver operating characteristic curve value was greater than 0.8, indicating high discrimination ability. For the PA assessment, the features that were most influential included the estimated glomerular filtration rate (eGFR), fasting blood sugar (FBS) level, serum creatinine (SCr) level, aspartate aminotransferase level, and alanine aminotransferase level. For the type of PA, the most influential features were weight, height, hemoglobin, FBS level, and high-density lipoprotein cholesterol level. For the weekly frequency of PA, the most influential features were eGFR, SCr level, height, FBS level, and weight.
Conclusions:
The methodology used in this study can contribute to the provision of PA prescriptions using clinical biomarkers. Furthermore, it can be used as a supplementary tool for quantitatively measuring the effects of lifestyle and motivating patients to maintain healthy lifestyles.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.