Analysis of Supervised Learning Techniques for Cost Effective Disease Prediction Using Non-Clinical Parameters

The rising costs of healthcare services have forced many to undergo annual screening procedures to reduce the incidence of chronic diseases. Diabetes being chronic in nature is always screened in the so called “wellness programs”. Data mining models involving clinical parameters obtained from screening procedures have been successfully implemented for disease prediction. This study sheds light on the use of some novel non-clinical parameters to predict diabetes. This move can be successfully implemented to reduce the screening cost of patients for diseases. The proposed model gives the best results with Random Forest classifier for diabetic prediction.