SISTEM PAKAR DIAGNOSA PENYAKIT KUCING MENGGUNAKAN METODE BAYES

Abstract

ABSTRACTRapid advances in technology have accelerated the adoption ofexpert systems to support decision-making across domains,including veterinary health. In cat care, diagnosing disease isoften challenging because clinical signs are not always apparent,increasing the risk of inappropriate treatment. This study aims todesign and evaluate an expert system for diagnosing felinediseases using the Bayesian method. The system accepts symptominputs from users, then performs probabilistic inference over aset of disease hypotheses based on a predefined rule/knowledgebase. Diagnosis is computed using Bayes’ theorem, updating theposterior belief of each disease given the observed symptoms.Experimental evaluation was conducted on representative testcases with relevant symptom combinations. Results show thesystem produces a probability ranking of candidate diseases,where the highest value indicates the primary diagnosis. In onetest case, the system yielded a maximum probability of 35.36%for liver-fluke infection (cacing hati, P05), identifying it as thedominant suspected disease. These findings indicate that aBayesian approach is feasible for early screening of feline diseasesand for providing structured initial treatment suggestions,offering a more consistent process than conventional visualassessment. Future work should expand the knowledge base andconduct broader validation (e.g., real-world case studies andoverall accuracy metrics) to further improve diagnostic reliabilityand support use in clinical settings as well as by cat owners.Keywords: expert system; Bayes’ theorem; feline diseasediagnosis; probabilistic inference; knowledge base