Research on the Performance of Different Machine Learning Algorithms in Hepatitis C Diagnosis

Hepatitis <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{C}$</tex> virus infection is a global public health problem, and early and accurate diagnosis is of great significance. In this study, 11 machine learning algorithms were used to classify hepatitis C by integrating UCL hepatitis dataset. Through random search hyperparameter optimization and Leave-One-Out Cross-Validation, it is found that Logistic Regression and Linear Discriminant Analysis have the best comprehensive performance, Support Vector Machine and Random Forest sample differentiation ability is strong, but KNearest Neighbor and Naive Bayes generalization ability is weak. New AST/ALT features improve model performance. The study pointed out that the data set category imbalance, hyperparameter optimization methods need to be improved and other problems, to provide a reference for the accurate diagnosis of hepatitis <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$C$</tex>, and help promote the development of clinical accurate diagnosis and public health prevention and control strategies.