Prediction of Heart Diseases and Cancer in Diabetic Patients Using Data Mining Techniques

Background: The heterogeneous, chronic diseases like heart diseases and cancer are commonly occur and increased nowadays in diabetic patients. Most of the people do not know the symptoms of these diseases and its chronic complications. Objective: The aim of this paper is to predict the diseases such as heart diseases and cancer in diabetic patients. The association between these diseases can be analyzed based on the factors that cause these diseases which include obesity, age, associated diabetic duration, and some other life style factors. Methods: This work consists of two stages. In the first stage, the attributes are identified and extracted using Particle Swarm Optimization (PSO) algorithm. In the second stage, ANFIS (Adaptive Neuro Fuzzy Inference System) with Adaptive Group based K-Nearest Neighbor (AGKNN) algorithm has been used to classify the data. Findings: The experimental results show a very good accuracy and signify the ANFIS with AGKNN along with feature subset selection using PSO. The performance is evaluated using performance metrics and proved this classifiers efficiency for the prediction of heart disease and cancer in diabetic patients. Application/ Improvement: This work demonstrates the diagnosis of diseases and its importance to predict it earlier. In future it can be implemented for other related diseases in medical data mining and healthcare.