Prognostic prediction of bilharziasis-related bladder cancer by neuro-fuzzy classifier

Cancer prognostic prediction requires a classification system that is robust to the interaction and uncertainty of input factors, as well as being interpretable on the decision made. In this paper, a hybrid neuro-fuzzy classifier is applied to determine the long-term outcome of patients with bilharziasis-related bladder cancer. The same data set is also analysed by a multi-layer perception neural network (MLPNN) and logistic regression, which are both widely used in the area of medical decision-making. In order to better assess the value of this neuro-fuzzy classifier, a benchmark data set used in this area of oncology, the Wisconsin breast cancer data (WBCD), is examined by the above three methods. The study demonstrates that the hybrid neuro-fuzzy classifier is efficient in cancer data analysis and it yields a high classification rate of 97.1% for WBCD, and 84.9% for the bladder cancer data, respectively.