Predicting Cardiovascular Risk in Athletes: Resampling Improves Classification Performance.

Area of Science:

• Sports Medicine
• Data Science
• Cardiology

Background:

• Cardiovascular diseases (CVDs) are a leading global cause of mortality.
• Athletes require medical clearance for competitive sports, necessitating risk assessment.
• Evaluating cardiovascular risk in athletes is crucial for public health.

Purpose of the Study:

• To evaluate a data mining methodology for cardiovascular risk assessment in athletes.
• To determine if the methodology can support clinical decision-making.
• To optimize the identification of at-risk individuals within an athletic population.

Main Methods:

• Collected anthropometric, demographic, and biomedical data from 26,002 athletes.
• Employed resampling techniques to balance class distribution.
• Utilized decision tree and logistic regression for risk classification.
• Assessed performance using receiver operating characteristic (ROC) curves.

Main Results:

• Data mining and resampling enhanced cardiovascular risk assessment accuracy.
• The area under the ROC curve (AUC) was improved by the methodology.
• Classification models demonstrated effectiveness in identifying at-risk athletes.

Conclusions:

• The proposed data mining methodology effectively assesses cardiovascular risk in athletes.
• This approach optimizes clinical decision-making and minimizes unnecessary examinations.
• The findings support the integration of data mining in sports medicine for proactive health management.