Prediction of Hepatitis disease using ensemble learning methods.

Abstract

Objective: Hepatitis is one of the chronic diseases that can lead to liver cirrhosis and hepatocellular carcinoma, which cause deaths around the world. Hence, early diagnosis is needed to control, treat, and reduce the effects of this disease. This study's main goal was to compare the performance of traditional and ensemble learning methods for predicting hepatitis B virus (HBV), and hepatitis C virus (HCV). Also, important variables related to HBV and HCV were identified.

Methods: This case-control study was conducted in Hamadan Province, in the west of Iran, between 2014 to 2019. It included 534 subjects (267 cases and 267 controls). The bagging, random forest, AdaBoost, and logistic regression were used for predicting HBV and HCV. These methods' performance was evaluated using accuracy.

Results: According to the results, the accuracy of bagging, random forest, Adaboost, and logistic regression were 0.65 ± 0.03, 0.66 ± 0.03, 0.62 ± 0.04, and 0.64 ± 0.03, respectively, with random forest showing the best performance for predicting HBV. This method showed that ALT was the most important variable for predicting HBV. The the accuracy of random forest was 0.77±0.03 for predicting HCV. Also, the random forest showed that the order of variable importance has belonged to AST, ALT, and age for predicting HCV.

Conclusion: This study showed that random forest performed better than other methods for predicting HBV and HCV.