Liver Disease Prediction Using Classification Algorithms

Abstract

Recently, Information Systems and strategic tools are being incorporated as additional means to aid the process of diagnosis of diseases in medical research. The liver, an essential organ is crucial in enzyme activation, bile production, metabolism of fats and storage of vitamins, glycogen and minerals. Liver diseases are difficult to diagnose and hence are often neglected due to the lack of proper symptoms at the initial stages. One of the most common symptoms of most liver diseases is hyperbilirubinemia which is hard to distinguish in early determination. In any case, this isn't quite certain and the perception of enzyme level is required to distinguish and affirm the nearness of liver illness.. Various machine learning techniques have been used in the prediction of liver diseases. In this research, we propose the usage of Decision Tree, Random Forest Algorithm and Support Vector Machine techniques in the prediction of liver disease by Binary Classification of the dataset into two given categories of patient experiencing liver sickness or not. The dataset contains information about patient attributes such as Total Bilirubin, Alanine Aminotransferase, Direct Bilirubin, Aspartate Aminotransferase, Age, Gender, Albumin, Total Proteins, Alkaline Phosphatase, Albumin and Globulin Ratio and the Result. The prediction from the above mentioned algorithms are compared on the parameters of Accuracy and various error calculations to determine the best suited algorithm.