A Literature Review in Data Mining Models Used for Survivability Prediction of Cancer Patients

Abstract

The research in the medical domain is clinical in its nature but with the advancement of information technology the trends of researchers in health care sector has been moving towards medical informatics. Usage of data mining techniques plays a major role in medical informatics. Especially when it comes to predicting or forecasting the survivability of a disease which is known as medical prognosis, data mining plays a major role. With the time medical prognosis is becoming highly important to increase the morality of patients especially who are diagnosed as cancer victims. Although the importance is increasing in the cancer prognosis, the methods that are in the practice for predicting still need to be improved and refined. In this paper we present an overview of the current research being carried out using the data mining techniques for prognosis of cancers. The goal of this study is to identify the well-performing data mining algorithms used on medical databases in order to predict survivability of cancer patients. The following algorithms have been identified: Decision Trees, Support Vector Machine, Artificial Neural Networks, Naïve Bayes and Fuzzy Rules. Analyses show that it is very difficult to name a single data mining algorithm as the most suitable for cancer prognosis. At times some algorithms perform better than others, but there are cases when a combination of the best properties of some of the aforementioned algorithms together results more effective.