Application of Data Mining to Predict the Status of PreART Patient

In Ethiopia, there are many people who live with HIV and started ART. There is also large amount of data on HIV patients and ART medications. The objective of this research is to predict Pre-ART patient status by applying DM techniques using CD4 count that helps the organization to better patient management, decision making, planning and better manage the disease. Business understanding and data preparation is conducted to know what the business indeed and what type of data the business process use. Besides data preprocessing is applied to prepare good quality datasets for experimentation. Guiding the overall knowledge discovery process CRISP-DM methodology was employed. Experimentation with neural network and J48 algorithm were implemented to build classifier model. J48 model is built with 10-fold cross validation testing technique with minNumObj=25 and ConfidenceFactors=0.25 and results the highest scores of 75.832%. Due to the quality of the data the maximum result of 75.832% is achieved. The researcher believed that the center has to strongly work on data recording, organization and management.