Personalized fitness recommendations using machine learning for optimized national health strategy
Year: 2025
Type: article
Abstract: Rising concerns over public health and chronic disease prevalence have intensified the demand for data-driven, personalized fitness interventions. While national health programs offer general guidelines, they often lack the granularity required to address individual variability in health status, lifestyle, and demographic context. This paper presents a machine learning framework to generate personalized fitness recommendations aligned with national health goals. Leveraging population-scale data, the aim is to optimize physical activity planning while maintaining fairness and clinical relevance across demographic subgroups. The study utilizes the National Health and Nutrition Examination Survey (NHANES) dataset, integrating biometric, behavioral, and demographic features. To enhance the behavioral relevance of our predictions, we integrated supplemental variables from the Behavioral Risk Factor Surveillance System (BRFSS), capturing psychological, motivational, and environmental factors that influence physical activity adherence. After preprocessing, models were developed using XGBoost, Decision Trees, and Artificial Neural Networks. Both regression (to estimate weekly activity minutes) and classification (to assign risk groups) tasks were addressed. Performance was evaluated through MeanIoU, Dice Score, sensitivity, and specificity. Demographic fairness was assessed via subgroup residuals and fairness gap analysis. XGBoost achieved superior performance, with a MeanIoU of 0.789 and F1 scores exceeding 0.79 across all risk categories. Model consistency was observed across age, gender, and ethnicity, with fairness gaps below 0.05. Residual error analysis and risk classification confirmed high reliability and low variance. The proposed system demonstrates the feasibility of using AI to personalize fitness plans at scale. It offers a pathway to integrate precision fitness with national policy, supporting equitable and effective public health strategies.
Source: Scientific Reports
Authors: Juan Chen, Yan Wang
Institution: Guangzhou Huashang College
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W4416574843