A novel holistic disease prediction tool using best fit data mining techniques

Abstract

Given that, today, the healthcare ecosystem is an information rich industry, there is an increasing demand for data mining (DM) tools to improve the quantity and quality of delivered healthcare; especially in handling patients suffering from deadly diseases such as HIV, Breast Cancer, Diabetes, Tuberculosis (TB), Heart diseases and Liver disorder. Given the fatality nature of these diseases when they remain undetected until at advanced stages, there remains a demand for best classifier tools to assist in diagnosing, detecting and treatment of these life-threatening diseases at their early stages. Complementary to this demand is the fact that the healthcare industry today generates large amounts of complex data about patients, hospital resources and disease diagnosis. Consequently, the healthcare ecosystem is warehousing large amount of medical data, which is an asset for healthcare organizations if properly utilized. The large amount of patient and disease related data could be processed and analyzed for knowledge extraction that enables support for cost savings and decision making towards delivery of timely and quality healthcare. In this paper, we report on an ongoing research work to develop and test a holistic DM disease prediction (Diagnosis and prognosis) tool, equipped with processes for preprocessing patients' data and a learning procedure for selecting a disease-specific best classifier, for disease prediction and delivery of speedy and cost effective diagnostic interventions and patient follow up in a hospital environment. As diseases are diagnosed, the predictive tool helps medical doctors in decision-making about what disease case it is and suggests possible treatment strategies within a much-reduced time. Test results for breast cancer and HIV data sets are reported. Achieved from the reported work are classification accuracies of 97.0752% (Classifier acting singly); 97.6323% (fusion of three classifiers). These results are better than those reported in the literature. The results show that the proposed DM disease prediction tool has potential to greatly impact on current patient management, care and future interventions against deadly diseases.