Experiments with Machine Learning in the Prediction of Coronary Artery Disease Progression

摘要: Fourteen classifiers were applied to the problem of coronary artery disease progression. The taken from different paradigms machine learning (symbolic, statistical and neural) in order encapsulate approaches. unsolved progression consists predicting stenosis (narrowing artery) change on basis clinical, laboratory epidemiological attributes. A total 263 patients belonging two classes (stenosis changed vs. non-changed) described with 25 overall results are not promising suggest that attributes used sufficiently relevant enable prediction It should also be pointed out simplest (the naive Bayesian classifier linear discriminant method) generally yield best results. This phenomenon seems typical for medical data is consistent our previous experience.