Predicting Access to Healthcare Using Data Mining Techniques

Healthcare is a very basic need for everyone in today's society. However, many individuals do not have or have difficulties accessing healthcare services. Researchers have studied various aspects in healthcare, including using data mining techniques to analyze healthcare data. Most of these studies, however, focused on treatment effectiveness, healthcare management, fraud and abuse detection, etc., few have reported on the accessibility study of healthcare. In this paper, we examine individual demand of receiving health care using the data from the Behavioral Risk Factor Surveillance System, a sample of the U.S. population. Models were built using the typical predictive modeling methods to predict the accessibility of healthcare based on the explanatory variables in the data set. Our experimental results showed that the regression and neural network models yielded better prediction accuracy whereas the decision tree and nearest-neighbor models fell behind in their performance.