Data Mining Approach for Predicting the Likelihood of Infertility in Nigerian Women

Abstract

According to WHO, there are 60 - 80 million infertile couples worldwide with the highest incidence in some regions of Sub-Saharan Africa. The social stigma of infertility weighs especially heavily on women, who bear the sole blame for barren marriages in many developing countries and may face divorce as a result. Interviews were conducted with gynecologists at one of the Teaching Hospitals in Nigeria in order to identify likelihood variables for infertility. 14 risk factors were identified and data collected from 39 patients from the hospital was pre-processed and the variables used to formulate the predictive model for the likelihood of infertility in women using three different decision trees algorithms. The predictive model was simulated using WEKA environment. The results revealed that C4.5 algorithm had the highest accuracy of 74.4% while the least performance was for the random tree algorithm with a value of 53.8%. This chapter presents a predictive model which can assist gynecologists in making more objective decisions concerning infertility likelihood.