relation: https://khub.utp.edu.my/scholars/5144/
title: Comparative assessment of rule-based and Bayes' theorem as inference engines in diagnosing symptoms for Islamic medication expert system
creator: Daud, H.
creator: Razali, R.
creator: Low, T.J.
creator: Sabdin, M.
creator: Zafrul, S.Z.M.
description: An expert system for diagnosing sickness and suggesting treatment based on Islamic Medication (IM) was constructed using Rule Based (RB) and Bayes' theorem (BT) algorithms independently as its inference engine. Comparative assessment on the quality of diagnosing based on symptoms provided by users for certain type of sickness using RB and BT reasoning that lead to the suggested treatment (based on IM) are discussed. Both approaches are found to be useful, each has its own advantages and disadvantages. Major difference of the two algorithms is the selection of symptoms during the diagnosing process. For BT, likely combinations of symptoms need to be classified for each sickness before the diagnosing process. This eliminates any irrelevant sickness based on the combination of symptoms provided by user and combination of symptoms that is unlikely. This is not the case for RB, it will diagnose the sickness as long as one the symptoms is related to the sickness regardless of unlikely combination. Few tests have been carried out using combinations of symptoms for same sickness to investigate their diagnosing accuracy in percentage. BT gives more promising diagnosing results compared to RB for each sickness that comes with common symptoms. © 2014 AIP Publishing LLC.
publisher: American Institute of Physics Inc.
date: 2014
type: Conference or Workshop Item
type: PeerReviewed
identifier:   Daud, H. and Razali, R. and Low, T.J. and Sabdin, M. and Zafrul, S.Z.M.  (2014) Comparative assessment of rule-based and Bayes' theorem as inference engines in diagnosing symptoms for Islamic medication expert system.  In: UNSPECIFIED.     
relation: https://www.scopus.com/inward/record.uri?eid=2-s2.0-84904131462&doi=10.1063%2f1.4882626&partnerID=40&md5=7fae913832cd67de28244d174ac370c6
relation: 10.1063/1.4882626
identifier: 10.1063/1.4882626