The liver is one of the most crucial organs in the human body. It performs several processes among them are metabolism, detoxification, bile formation, storage and blood-management, immunological function. Hepatitis, fatty liver disease, cirrhosis, and liver cancer are examples of the illnesses that can dangerously affect the liver. A liver transplant can be essential if the liver is seriously damaged or is not working properly. Liver function can be evaluated by diagnostic testing. The condition known as cirrhosis is a late stage of liver scarring (fibrosis) brought on by a variety of liver illnesses and disorders, including chronic hepatitis, alcoholism, fatty liver disease, autoimmune hepatitis, and a few genetic liver diseases. In addition to a physical examination, medical history, blood tests, imaging tests (such as an ultrasound, CT scan, or MRI), and occasionally a liver biopsy, cirrhosis is diagnosed. In this paper, a machine learning based model is used in order to detect, classify and predict the degree of cirrhosis based on previous regular laboratory tests only. Liver cirrhosis is classified into 3 classes: (F0-F1) for normal liver, (F2) for a moderate stage of liver cirrhosis, and (F3-F4) for complete liver cirrhosis. The algorithms used in this study are support vector machines, artificial neural networks, Gradient Boosting, K-Nearest Neighbor, and Naive Bayes. Results showed that, the Gradient Boosting algorithm achieved the best performance during both learning and testing phases with accuracy level of 86% during learning and 100% during testing.

Machine learning Liver cirrhosis liver laboratory tests.