Comparative Analysis of Machine Learning and Deep Learning Techniques for Liver Disease Prediction

Liver cancer is one of the world’s largest causes of deaths to humans. The American Cancer society estimated that throughout the world more than 700,000 people are diagnosed with liver cancer every year and 600,000 deaths occur each year. Liver is an important organ performing more than 12 functions including purification of blood, production of bile, excretion of waste and so on. In recent years, liver diseases are the cause for more number of deaths globally. Predicting the disease at earlier stage will be helpful in reducing the mortality rate. But early prediction is difficult since the symptoms of liver malfunction are recognized only at later stages of the disease. This paper presents a comparative study on the various machine learning techniques like SVM, regression, decision tree, K-NN, Random Forest, etc., and deep learning techniques like CNN, LSTM, U-Net and transfer learning. Above techniques are evaluated and compared in terms of precision, Recall, F1-score, accuracy. Various liver disorders covered in this study include cirrhosis, hepatitis, liver tumors, liver lesions and liver cancer. The comparative study aims to choose the best technique for early prediction of liver disease. The experimental results show that deep learning techniques outperform more than machine learning techniques.