SOURCE:

Faculty: Engineering
Department: Electronic And Computer Engineering

CONTRIBUTORS:

Osigbemeh, M. S.
Inyiama, H.C.

ABSTRACT:

Adevelopment of a fuzzy neural Expert System for contagious disease detection and isolation was achieved in this research. The back propagation algorithm (BPA)was used to establish a direct mathematical connection with the vague and ambiguous reporting of patient’s symptoms as manifested during pre-diagnosis to help identify and initiate quarantine procedures for index cases of contagious diseases. The adopted methodology involved a Fuzzy preprocessing of the reported symptoms toreduce nuances, generate normalized and fuzzified valueswhich then serves as inputs to the supervised back propagation artificial neural network that executes a gradient descent optimization scheme. The BPA was found to preserve and correlate the results of the fuzzy inferencing system during simulationwith 95.87% of both the 20 cases (10 confirmed and 10 suspected) of validation data on Lassa fever obtained from Federal Teaching Hospital, Abakaliki in Ebonyi State, Southeast Nigeria and 14 cases (7 confirmed and 7 probable) from the Institute of Lassa Fever Research and Control at Irrua Specialist Teaching Hospitalin Edo State, Nigeria. Validation tests based on demographic data onsymptoms in confirmed and probable Ebola cases of1151 patients from Guinea, Liberia, Nigeria, and Sierra Leone from the WHO Ebola Response Teamshowed 95% convergence rate. The percentage was obtained by considering the results of the output convergence in all presented symptoms to the ANN at a bench mark of 620 epochs. An average False Alarm Ratio of 5.37% of both data was obtained with the error traceable to incomplete presented data during the several iterations in the network.The k-nearest neighbor algorithm provided an instance based learning scheme for classifying and estimating new cases.This work which fulfills an identified need in modern healthcare safety delivery systems is recommended for deployment and use in standalone interfaces in hospitals and clinics.