A Novel Statistical Diagnosis of Clinical Data

In this paper, we present a diagnosis method of diseases from clinical data. The data are routine test such as urine test, hematology, chemistries etc. Though those tests have been done for people who check in medical institutes, how each item of the data interacts each other and which combination of them cause a disease are neither understood nor studied well. Here we attack the practically important problem by putting the data into mathematical setup and applying support vector machine. Finally we present simulation results for fatty liver, gastritis etc and discuss about their implications.