Optimizing physical education schedules for long-term health benefits

Abstract

Introduction: Physical education (PE) plays a vital role in promoting long-term health and wellness among students. Effective scheduling of PE classes is essential for maximizing fitness improvements across diverse populations. However, traditional approaches to optimizing PE schedules may not adequately account for individual differences in demographics and activity patterns.

Methods: This study proposes an efficient method for optimizing PE schedules using deep learning (DL) techniques. The developed DL model integrates convolutional neural network (CNN) layers to capture spatial features and long short-term memory (LSTM) layers to extract temporal patterns from demographic and activity-related variables. These features are combined through a fusion layer, and a customized loss function is employed to accurately predict fitness scores.

Results: Extensive experimental evaluation demonstrates that the proposed model consistently outperforms competitive baseline models. Specifically, the model achieved notable improvements in mean squared error (MSE) by 1.35%, R-squared R ² by 1.18%, and mean absolute error (MAE) by 1.22% compared to existing approaches.

Discussion: The findings indicate that the DL-based approach provides an effective method for optimizing PE schedules; resulting in increased fitness levels and potential long-term health benefits. This model can assist educational institutions and policymakers in designing and implementing effective PE programs personalized to diverse student populations.

Keywords: data-driven public health; fitness score prediction; health promotion strategies; long-term health benefits; physical education.