What We Found on Our Way to Building a Classifier: A Critical Analysis of the AHA Screening Questionnaire

The American Heart Association (AHA) has recommended a 12element questionnaire for pre-participation screening of athletes, in order to reduce and hopefully prevent sudden cardiac death in young athletes. This screening procedure is widely used throughout the United States, but its efficacy for discriminating Normal from Non-normal heart condition is unclear. As part of a larger study on cardiovascular disorders in young athletes, we set out to train machine-learning-based classifiers to automatically categorize athletes into risklevels based on their answers to the AHA-questionnaire. We also conducted information-based and probabilistic analysis of each question to identify the ones that may best predict athletes’ heart condition. However, surprisingly, the results indicate that the AHA-recommended screening procedure itself does not effectively distinguish between Normal and Non-normal heart as identified by cardiologists using Electroand Echo-cardiogram examinations. Our results suggest that ECG and Echo, rather than the questionnaire, should be considered for screening young athletes.