This work presents the application of machine learning techniques to analyze the influence of physical exercise in the heart's physiological properties, during ventricular fibrillation. With that purpose, different kinds of classifiers (linear and neural models) were used to classify between trained and sedentary rabbit hearts. These classifiers were used to perform knowledge extraction through a wrapper feature selection algorithm. The obtained results showed the higher performance of the neural models compared to the linear classifier (higher performance measures and higher dimensionality reduction). The most relevant features to describe the benefits of physical exercise are those related to myocardial heterogeneity, mean activation rate and activation complexity.