Diagnosis of liver disease with machine learning methods

Liver diseases are among quite common diseases seen worldwide. Early and accurate diagnosis of these diseases has vital importance due to their very frequent occurrence and high mortality rate. Traditional diagnostic methods are still used in medicine. However, today, thanks to the developing artificial intelligence technologies, powerful tools can be provided to support physicians in disease diagnosis, detection and treatment processes. In this thesis, Machine learning algorithms, which is a subfield of artificial intelligence have been used for the diagnosis of liver diseases. In this study, using WEKA data mining tool by means of J48, Logistic Model Tree (LMT), Decision Stump, Hoeffding Tree, REP Tree, Random Forest, Random Tree and IBk machine learning algorithms are studied on Liver Patient Data Set (ILPD). The aim is to achieve the best diagnostic result with these algorithms. Data preprocessing processes different from literature were also used in the study. In the first part of the study, emphasizing the importance of liver diseases and machine learning, similar studies in the literature were included. In the second part, theoretical foundations related to machine learning and data mining were included. In the third section, detailed information was given about the data set used, the algorithms in the application were introduced, and how the data was passed through preprocessing and used with the algorithms in the WEKA tool was explained. In the fourth chapter, the universal classification performances of the obtained models are analyzed according to the evaluation criteria. In the fifth and final section, the results obtained from the analyzes were evaluated and interpreted. The best classification models attained from machine learning algorithms have the potential to create inference mechanisms of intelligent systems to be developed for Disease Diagnosis and treatments. In the light of the information obtained from this thesis, this study can be transformed into an intelligent system that is included in hospital information management systems, can access laboratory results and perform liver cancer screening automatically and make necessary warnings to doctors about their patients. In the following process, it is aimed to transform this system foreseen into a comprehensive health project. This study is important for the diagnosis of liver diseases in order to assist physicians in the field of liver diseases with the classification algorithms used for early diagnosis and screening of liver diseases.