Performance Assessment of Classification Algorithms on Early Detection of Liver Syndrome

Area of Science:

• Medical Informatics
• Machine Learning
• Hepatology

Background:

• Liver disease is a growing global health concern, disproportionately affecting younger populations.
• Early detection of liver disease is challenging but critical, as liver failure has a rapid and severe prognosis.
• Machine learning approaches are being explored for early and accurate liver disease prediction.

Purpose of the Study:

• To compare the performance of ten different machine learning classifiers for early liver disease prediction.
• To identify the optimal machine learning model for accurate and timely diagnosis of liver conditions.

Main Methods:

• Ten classifiers were evaluated: A1DE, Naive Bayes (NB), Multilayer Perceptron (MLP), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), CHIRP, Classification and Regression Tree (CDT), Random Forest (RF), J48, and Forest-PA.
• Datasets were sourced from the UCI Machine Learning Repository and GitHub.
• Performance was assessed using metrics including Root Mean Squared Error (RMSE), Root Relative Squared Error (RRSE), recall, specificity, precision, G-measure, F-measure, Matthews Correlation Coefficient (MCC), and accuracy.

Main Results:

• Random Forest (RF) demonstrated superior performance on the UCI dataset, achieving an accuracy of 72.1739% with RMSE of 0.4328 and RRSE of 87.6766.
• Support Vector Machine (SVM) achieved the highest accuracy (71.3551%) on the GitHub dataset, outperforming other tested techniques.
• Both RF and SVM showed promising results, indicating their potential for early liver disease detection.

Conclusions:

• The study highlights the effectiveness of machine learning, specifically RF and SVM, in predicting liver disease.
• The findings provide a benchmark for future research aiming to improve liver disease prediction models.
• Further research can leverage these results to develop more robust and accurate diagnostic tools for liver conditions.