<![CDATA[The advancement of artificial intelligence technology presents new opportunities to support medical professionals in making faster and more accurate clinical decisions. This study introduces a liver disease classification system based on the Naive Bayes algorithm, designed to be easily interpretable by doctors and healthcare personnel. A dataset of 580 patients with 11 clinical attributes—ranging from bilirubin levels to albumin–globulin ratio—was used and processed through data cleaning and normalization stages. The Bernoulli Naive Bayes model was then trained and evaluated using a confusion matrix and ROC-AUC analysis. The results show an accuracy of 67%, with strong performance in identifying patients at risk of liver disease (recall of 0.82), but weaker in classifying healthy individuals (recall of 0.28). The fast training time and transparent probabilistic predictions of the Naive Bayes algorithm make it a practical solution for developing a prototype of a medical decision support system. Future recommendations include incorporating additional relevant clinical features and applying ensemble methods to improve diagnostic sensitivity and specificity.]]>
Copyrights © 2025
