Prediction of liver cancer development risk in genotype 4 hepatitis C patients using knowledge discovery modeling

Hepatitis C is a primary reason for the liver cancer, which is a leading cause of death. The objective of this paper is to predict the hepatitis C infection progression into cirrhosis or liver cancer. For the prediction of the disease progression, a knowledge discovery framework is proposed consisting of three phases: preprocessing, data mining and prediction. While the preprocessing phase focuses on the discretization of the training data, the data mining phase focuses on mining patients' records using a rule based classifier built by the proposed algorithm to generate a set of unique rules. Eventually, the predictor uses the rules to predict patients' disease progression. Experimentation on 1908 chronic hepatitis C Egyptian patients with 27 extracted features collected from blood samples were used to train the model, with other 406 patients' cases for testing which showed accuracy 99.5 %.