A decision support system for cancer prevalence in South Africa

Cancer is one of the terminal diseases humanity has to deal with today. It affects both old and young, male and female, all races and regions. In this paper, we implement a software system that monitors cancer incidences in Eastern Cape Province of South Africa. Using a Bayesian network model, the system predicts the likelihood of getting a particular type of Cancer taking into account the causal relationships, which may be gender, race, location of residence, and age group among the variables considered in this work. We collected data of Cancer patients from some of the biggest hospitals in Eastern Cape Province, aggregated, classified and analyzed them using Bayesian network model. The preliminary results show that accurate prevalence rates and prediction of getting any type of Cancer in Eastern Cape Province while residing in any of the districts can be trusted since they are based on actual data. The results of this work can be used by people in deciding where to reside in Eastern Cape Province. The South African government can also utilize it to sensitize people of the prevalence by conducting campaigns in areas mostly affected. We successfully correlated data and computed causal relationship among the variables considered. We attained a prediction rate of 97% accuracy.