Optimization of Certainty Factor in Cat Disease Diagnosis Using Particle Swarm Optimization

The increasing prevalence of diseases in pet cats highlights the need for a diagnostic support system that is both fast and accurate. The Certainty Factor (CF) method is widely used in expert systems to represent uncertainty in decision-making. However, its reliance on subjective expert judgment can reduce diagnostic accuracy and system confidence. This study aims to optimize CF values using the Particle Swarm Optimization (PSO) algorithm, enabling the system to adapt better to actual medical data. CF values were initially collected from two veterinary experts and combined using the median method to minimize bias. These values were then optimized using PSO, with parameter tuning performed individually for each disease to maximize fitness. The dataset used for validation consisted of 100 medical records of cats diagnosed with one of nine common feline diseases, including Ringworm, Scabies, Helminthiasis, and others. The results show an increase in diagnostic accuracy from 85% (using original CF) to 88% (after PSO optimization). Moreover, 70.73% of the cases with consistent diagnoses before and after optimization showed an increase in final CF values, indicating greater confidence in the system’s diagnostic decisions. These findings suggest that the integration of CF with PSO not only improves diagnostic accuracy but also strengthens the reliability of expert systems in the veterinary field, particularly for early and efficient identification of cat diseases.