DECISION SUPPORT DIAGNOSIS MODELING USING ARTIFICIAL NEURAL NETWORK AND FUZZY LOGIC TECHNIQUES IN CASE OF MALARIA

Abstract

Malaria in sub Saharan countries especially in Ethiopian is the major one killer disease that can’t pushed mainly in lowland areas. In Ethiopia the major challenge for healthcare service diagnosis is shortage of skilled manpower. The number of health professionals and patients’ demand are disproportionate according to Ethiopian ministry of health and WHO report. As a result several lives have been lost while others are living with deteriorated health status. This research study had developed artificial neural network and fuzzy logic soft computing techniques which provides an efficient means of handling the complexity associated with the diagnosis of malaria. Both techniques investigated to develop an intelligent decision support diagnosis to improve the ability of physician. The study built artificial neural network (ANN) model which helps classifies malaria patterns and fuzzy logic (FL) model which helps decision diagnosis support prediction. The ANN model is a supervised learning techniques uses for classification of malaria dataset to their classification level. A fuzzy logic model was developed using mamandi inferences engine techniques to decide a decision making prediction process. The feature value of malaria dataset diagnosis serve as the core input parameters to the ANN and FL. Experimental study of the research study was conducted using medical records of malaria patients from Arba Minch Referral hospital. The performance of the ANN study were evaluated using cross entropy, ROC and confusion matrix. Fuzzy logic model was defuzzified using centroid of gravity defuzzifier.