Web-Based Decision Support System for Liver Fibrosis Prediction Using Combination of Principal Component Analysis and Artificial Neural Network

Abstract

Hepatitis has become one of the diseases that contribute to many mortality rates. Hepatitis patients are at high risk of getting Liver Fibrosis as their tissue is already in inflammation. A condition of liver fibrosis itself is where the tissue organs are producing excessive protein including collagen. This condition is worsened as the tissue are failed to regenerate and causing the blockage of the blood supply into the liver. In order to get the best treatment and prevent liver fibrosis continue to an advanced stage, early diagnosis is required. This study proposed a web-based decision support system using a combination of principal component analysis and an artificial neural network. This system is planned to classify liver fibrosis. From the principal component analysis part, which are PC-20, PC-23, and PC-27. These principal components are calculated to get better classification. And then the step continued in the artificial neural network (ANN). The best topology used in this ANN shows a classification accuracy of 96%. This application is designed web-based. It also tested using User-Centered Design to measure if the application is fit in the user’s point of view. In conclusion, this web application has usability level of “GOOD” and 0.74 usability points.