Constructing a novel mortality prediction model with Bayes theorem and genetic algorithm

摘要: Intensive care is one of the most important components modern medical system. Healthcare professionals need to utilize intensive resources effectively. Mortality prediction models help physicians decide which patients require and do not. The Simplified Acute Physiology System 2nd version (SAPS II) popular mortality scoring systems currently available. This study retrospectively collected data on 496 admitted units from year 2000 2001. average patient age was 59.96+/-1.83years old 23.8% died before discharge. We used these as training constructed an exponential Bayesian model by combining BSM (Bayesian statistical model) GA (genetic algorithm). optimal weights parameters were determined with GA. Furthermore, we prospectively 142 for testing new model. this group 57.80+/-3.33years 21.8% power better than SAPS II (p<0.001). can manage both binary continuous data. rate predicted be high if patient's Glasgow coma score less 5.