Use of data mining to check the prevalence of prostate cancer: Case of Nairobi County

Prostate cancer has been on the rise in the past years and alarming cases being found in men in their 20's. The problem is that most of the cases are diagnosed in their late stages thus the mortality rate being high. In recent years data driven analytic studies have become a common complement with novel research where different tools and algorithms are taking a centre stage in cancer research. In this research, the main objective was to use data mining to derive patterns which were used in building a prognostic tool that helps in identification of the Gleason score once screened and deciding the treatment technique. In this research, we used two popular data mining tools (R Environment and WEKA) which exhibited almost same results. The dataset contained around 485 records and 7 variables. In WEKA, a 10-fold cross-validation was used in model building and comparison between ANN and J48. The results showed that ANN is the most accurate predictor compared to J48 in all the instances displaying varying levels in the different zones created. This study contributes to society, academics and cancer research which ultimately assist in reduction of mortality rates by use of pattern recognitions which leads to better decision making. Furthermore, this is a potential impact in helping the GOK (Government of Kenya) in establishing where they should correctly place the cancer diagnosis and treatment equipment that were rolled out by the National government early 2015.