Liver Disease Prediction Using Machine Learning Techniques

Abstract

The problem of liver disease is considered a crucial one in global public health which is usually diagnosed only at advanced stages causing serious health issues and even death. The key problem under consideration in this project is the insufficient efficiency of liver disease diagnosis methods. The aim of this project is to create a predictive system based on machine learning methods which will be able to classify the patients as either having liver disease or not depending on the provided clinical information. In order to reach the stated aim, we have used the dataset that contains information about 583 patients with the biochemical features. Data preprocessing operations like dealing with missing data, feature encoding and normalization were conducted to increase data quality. We have developed several classifiers like Random Forest, AdaBoost, Support Vector Machine and Naive Bayes Classifier. The best results among them were obtained by Random Forest algorithm with the accuracy of 77.8%.