Liver Disease Prediction Using Machine Learning Classification Techniques

Abstract

Machine Learning enables the discovery of patterns
in large datasets, facilitating decision-making by
allowing machines to undergo learning processes
(supervised, unsupervised, semi-supervised, or
reinforced). This study utilizes a dataset of liver
patients from the UCI Repository, employing
supervised learning techniques. The dataset
comprises extensive information from medical
examinations of liver patients, which can be utilized
to improve their future conditions. Historical and
categorized patient data serve as input for various
algorithms to predict future patient outcomes. The
algorithms used in this study for liver patient
prediction include Logistic Regression, Decision
Tree, Random Forest, k-Nearest Neighbors, Gradient
Boosting, Extreme Gradient Boosting, and
LightGBM. Analysis and results indicate that these
algorithms achieve high accuracy following feature
selection.