LIVER DISEASE PREDICTION USING MACHINE LEARNING ALGORITHMS

Abstract

A common illness besides heart attacks, which claim numerous lives, is liver disease. Due to the fact that liver disease is commonly discovered at a late stage, which results in death. Numerous factors that may affect health parameters, such as excessive alcohol consumption, breathing harmful gases, consuming polluted water, the absorption of poisoned nutrients, consuming treated foods without necessity, ingesting drugs, and others, are contributing to the rise in the number of liver patients year by year. These health parameters can be used in the early stages of machine learning prediction models to predict liver diseases. A lack of detection may result in improper medication and medical care. In order to aid the medical professional in prescribing the right medications and providing medical care, accurate detection is required. Some research teams have used data collection strategies to identify liver disease. The issue is that reaching a consensus on a better classification algorithm for liver disease is difficult. This study aims to determine whether the patient has liver disease problems using a data set obtained from the UCI Machine Learning repository that contains 583 instances and 10 key characteristics. Classification algorithms such as Logistic Regression, KNN algorithm, the SVM classifier, Naive Bayes, Decision Tree algorithm, and Random Forest are applied to this data set. These algorithms are compared using the accuracy metric. The best-chosen algorithm among these is Logistic Regression as it provides the best accuracy, i.e., 73.28%.