Non -Alcoholic Fatty Liver Early Prediction Using Squared Reinforcement Learning Based on Primary Biliary Cirrhosis

Abstract

Biomedical engineering and artificial intelligence encompass many intricate challenges worthy of investigation. An essential challenge within this domain is identifying an effective classifier algorithm for predictive analytics. This challenge holds significant importance, primarily due to the extensive time it often requires for resolution. Hence, developing an algorithm that autonomously identifies optimal classification methods becomes vital. Classification algorithms play a pivotal role in diagnosing and forecasting various health-related issues. Leveraging artificial intelligence to predict human diseases is particularly beneficial among various applications. A prominent example of this application involves the prediction of primary biliary cirrhosis utilizing classification algorithms. This research presents a reinforcement learning framework designed to autonomously acquire the most effective classification and balancing algorithms for predicting this disease. The proposed framework draws inspiration from a voting mechanism. This approach is designated as Square Reinforcement Learning (SRL) by integrating four distinct classification metrics. The findings of this study demonstrate that the proposed SRL method enhanced the performance of classification algorithms, increasing accuracy from 85% to an impressive 98%.