A Cost Sensitive Ensemble Method for Medical Prediction

In this paper we present an analysis of the importance of cost sensitive methodology in medical disease prediction. And a C5.0 based cost sensitive ensemble approach is presented to support improving the effectiveness and decreasing the misdiagnosis cost in the same time. The verification is conducted on the real cases from Changhai Hospital Renal Registered (CHRR) data which consists of 483 records, and 25 fields are used through variable selection. The real case experiments show that, our approach works effectively and could be used as an assistant approach for peritonitis analysis in some circumstances.