A Unified Machine Learning Framework for Effective Prediction of liver disease – Fatty Liver Towards Cirrhosis

Abstract

Liver is the largest organ of the human body with more than 500 vital functions. In recent decade number of liver patient has been reported such as cirrhosis, fibrosis, or other liver disorder. There is a need of effective, early, and accurate identification of individual suffering with such disease, so that the person may recover before the disease spread and become a fatal. For this, applications of Machine Learning are playing a significant role. In this research there are two main sub activities has been performed, firstly a protocolbased literature review (SLR) has been done. Secondly based on the SLR a unified Machine learning framework named as Machine Learning Based Liver Disease Diagnose (MaLLiDD) has been formed. In SLR phase we have reviewed 44 articles extracted from 5 different electronic repositories published from January 2015 to November 2021. After a systematic and protocol-based review we answered 6 research questions about machine learning algorithms. The identification of effective feature selection technique, data imbalance management technique, accurate machine learning algorithms, list of available data sets with their URL’s and characteristics, and feature importance based on usage has been identified for diagnosing the liver disease. The reason to selecting this research question is, in any machine learning framework the role of dimensionality reduction, data imbalance management, Machine learning Algorithm with its accuracy and data itself is very significant. MaLLiDD is a unified framework that has been tested on 3 different datasets to diagnose fatty liver towards cirrhosis and its severity level. The main contribution of this research is the framework MaLLiDD (that has been formed based on results of 44 studies), that can work for three different datasets helps in diagnosing liver disease. Performance given by MaLLiDD on cirrhosis diagnoses data set is 99.8 % , on ILPD Indian liver patient dataset 76.5% accuracy has been obtained and 76.1 % of accuracy on cirrhosis staging (severity level ) dataset.