Development of fuzzy logic-base diagnosis expert system for typhoid fever

Abstract

Typhoid fever (TyF), caused by salmonella typhoid bacteria, represents one of the main public health challenge in various parts of the world. It is often treatable when diagnosed early, but if left untreated could lead to other medical complications. This study proposed an artificial intelligence means (arim) for diagnosis of TyF. The objectives are to find out the leading risk factors for TyF, develop fuzzy logic base-expert system, called Typhoid Responsive Expert System (TyRes), that can predict the ailment from symptoms and use TyRes to predict TyF in patients. Two sets of questionnaires were used for data collection. 325 copies were administered to the patients in 25 hospitals in Lagos, Abeokuta and Ifo, South-west Nigeria. Another set of 200 copies were administered to human medical experts (hme), 70 doctors and 140 qualified nurses, to capture hme knowledge about TyF and its symptoms. The data was analysed using Chi-Square to identify the main symptoms spotted by most of the hme. TyRes was implemented in Matlab 2015a using the main factors as input variables. Vomiting, high-temperature, weakness, abdominal-pains and loss-of-appetite were the input variables used to develop TyRes. When tested to predict TyF in 25 patients, 76% accuracy was derived when comparing hme predictions with TyRes results. It can be concluded that TyRes can mimic hme by 76% of all TyF predictions. The arim is considered reliable and can be used at home, school and health centres where hme are scarce.