Reliable diagnosis of acute abdominal pain with conformal prediction

摘要: Medical decision support is an area in which a lot of machine learning research has been conducted and several diagnostic prognostic systems have developed. The majority these only produce bare predictions, without any indication how reliable each predictions is. An this kind however, highly desirable especially the medical field. In paper we address problem with use recently developed technique, called conformal prediction, for accompanying traditional algorithms measures their accuracy reliability. We apply prediction based on Neural Network classifier to acute abdominal pain diagnosis obtain high level are complemented well-calibrated practically useful confidence measures.