Gradient Boosting Model Accurately Evaluates Exercise Outcomes in 220 College Students

A gradient boosting tree machine-learning model demonstrated strong predictive validity in evaluating college students’ physical exercise outcomes from combined physiological and behavioral tracking data, according to an eight-week empirical investigation tracking 220 undergraduates. The computational framework synthesized multidimensional variables into a continuous efficacy score, achieving a mean squared error of 0.82 and a coefficient of determination of 0.87. Researcher Qinfeng Dong reported that the data-centric system successfully identified weekly exercise frequency and heart rate fluctuation as the primary dominant factors governing overall training evaluation.

The investigation focused on resolving longstanding operational challenges in university sports pedagogy and physical education curricula. Traditional physical education assessments frequently depend on isolated performance metrics or single-time fitness tests. According to the research, such conventional evaluation practices fail to capture the continuous and dynamic interplay between multidimensional behavioral and physiological indicators, specifically cardiorespiratory adaptation, training frequency, and recovery efficiency. Consequently, this structural measurement limitation obstructs instructors and trainers from delivering timely, personalized feedback to college students seeking to improve their physical fitness and conditioning regimens.

To bridge this pedagogical and methodological gap, the researcher established an analytical framework powered by the gradient boosting tree algorithm. The system was designed to synthesize multi-sourced behavioral and physiological inputs into a unified, continuous efficacy score. During the eight-week monitoring period, the study tracked 220 undergraduate students across several specific health and lifestyle metrics, comprising weekly exercise sessions, heart rate variability, body mass index, and sleep duration. By integrating these diverse parameters into a single continuous score, the data-centric methodology aimed to convert subjective and empirical training adjustments into structured, evidence-based instructional decisions for university physical education programs.

Quantitative assessment of the gradient boosting tree model demonstrated high predictive accuracy across the tracked undergraduate cohort. The algorithm attained a mean squared error of 0.82 alongside a coefficient of determination of 0.87, confirming its strong predictive validity and capacity to evaluate exercise efficacy scores reliably. Analysis of internal feature importance revealed that weekly exercise frequency served as the dominant influencing factor in the predictive architecture, registering an importance metric of 0.32. Heart rate fluctuation constituted the second most influential variable, recording a measured importance metric of 0.25 within the multi-sourced evaluation model.

The resulting predictive framework establishes a quantitative and individualized feedback loop for university sports pedagogy and higher education physical education programs. According to the author, identifying exercise frequency and heart rate fluctuation as the primary drivers of physical exercise outcomes allows educators to design targeted training modifications. These targeted adjustments include optimizing the specific frequency and intensity of weekly exercise sessions for individual undergraduates. Through this data-centric approach, the gradient boosting tree framework provides a dynamic and holistic evaluation system to facilitate evidence-based program adjustments, replacing isolated traditional tests with continuous performance feedback.

Subject of Research: Machine learning evaluation of college student physical exercise outcomes

Article Title: Application of gradient boosting tree algorithm in quantitative evaluation and program adjustment of college students’ physical exercise outcomes

Article References: Dong, Q. (2026). Application of gradient boosting tree algorithm in quantitative evaluation and program adjustment of college students’ physical exercise outcomes. Discover Applied Sciences. https://doi.org/10.1007/s42452-026-09832-2

Image Credits: AI Generated

DOI: 10.1007/s42452-026-09832-2

Keywords: gradient boosting tree, physical education, exercise outcomes, heart rate variability, exercise frequency, heart rate fluctuation, body mass index, sleep duration, continuous efficacy score, predictive validity, undergraduate students, sports pedagogy, training adjustments, cardiorespiratory adaptation, recovery efficiency, individualized feedback, university athletics