Prediction of Cardiovascular Risk Using Machine Learning Based on Maximal Oxygen Consumption, Physical Fitness, and...

Area of Science:

• Sports Medicine
• Public Health
• Biostatistics

Background:

• Cardiovascular risk factors established in adolescence often persist into adulthood.
• Low cardiorespiratory fitness is a significant predictor of future cardiovascular disease.
• Early identification of cardiovascular risk in adolescents is crucial for preventative interventions.

Purpose of the Study:

• To classify cardiovascular risk in adolescents associated with low cardiorespiratory fitness.
• To develop and evaluate machine learning models using anthropometric and muscular fitness variables.
• To assess the feasibility of using school-based tests for cardiovascular risk screening.

Main Methods:

• Analysis of a cross-sectional dataset of 7,852 adolescents (aged 13-16 years).
• Predictors included BMI, waist circumference, waist-to-height ratio, standing long jump, push-ups, and sit-ups.
• Machine learning models (Gradient Boosting, Random Forest) were trained and evaluated using cross-validation and SMOTE for class imbalance.

Main Results:

• Ensemble tree-based models, particularly Gradient Boosting, showed superior predictive performance (AUC-ROC 0.716).
• Muscular fitness measures (push-ups, sit-ups, standing long jump) were the most important predictors.
• Anthropometric indicators had lower importance in classifying cardiovascular risk.

Conclusions:

• Machine learning models utilizing school-based muscular fitness tests and anthropometry can effectively discriminate adolescents at cardiovascular risk.
• This approach offers a feasible and low-cost strategy for early identification of at-risk youth.
• Findings support targeted physical activity interventions for adolescents with low cardiorespiratory fitness.