Establish VO₂max prediction models based on exercise and body parameters from the step test

This study addresses the challenge of cardiorespiratory fitness (CRF) assessment by proposing predictive models for maximal oxygen uptake (VO₂max) based on step test parameters. Recognizing VO₂max as a gold standard for CRF evaluation, this study aims to develop a VO₂max prediction model based on a step test, providing a simple and practical alternative for primary healthcare and health monitoring. This model enables clinicians and health management professionals to efficiently assess patients' cardiorespiratory fitness. Through the recruitment of 200 healthy Taiwanese adults, the research combined direct VO₂max measurements with step test heart rate (HR) data and variables like age, sex, percentage body fat (PBF), body mass index (BMI), and resting heart rate (RHR) to develop six predictive models. This method is applicable for clinical health monitoring, cardiorespiratory fitness assessment in patients with chronic diseases, and exercise capacity monitoring in cardiac rehabilitation programs. The study identified that PBF-based models consistently outperformed BMI-based ones, with Model^PBF3, which incorporates HR responses during exercise, achieving the highest accuracy (R² = 0.689; SEE = 4.6971 ml·kg⁻¹·min⁻¹). These results indicate that the model can effectively estimate VO₂max and be applied in primary healthcare, remote health monitoring, and cardiac rehabilitation settings, providing a simple and practical tool for cardiorespiratory fitness assessment in clinical practice. Validation via PRESS cross-validation and Bland-Altman plots confirmed the stability and reliability of the models across diverse subgroups. By bridging the gap between laboratory-grade precision and everyday practicality, the study introduces a robust, low-cost, and user-friendly tool for CRF assessment, adaptable for non-athletes and those unable to perform high-intensity exercises. This research advances the feasibility of CRF self-management in varied settings, while future iterations could extend its applicability to broader demographics and integrate additional physiological variables for universal adoption.