Predicting Liver Disorders Using an Extreme Learning Machine

Liver diseases are caused by excessive alcohol intake or viral infection. The liver can fail or develop cancer if it is not detected in its early stages. In the proposed approach, liver function tests are used to detect liver disease at an early stage. Liver disease datasets also suffer from the imbalance problem, just like various real-world datasets. In addition, some samples may be noisy and redundant, which can increase computing costs and degrade classification performance. The purpose of this study is to develop a mechanism that will assist medical practitioners in accurately and reliably predicting and diagnosing liver disease. This hybrid activation function-based extreme learning machine uses noise removal using a k-NN filter, redundant pair elimination, SMOTEEN as a data balancing technique, and a Hybrid Activation function in an extreme learning machine (HAELM). The experimental result of the proposed method of extreme learning machine demonstrated highly acceptable results over liver patient datasets.