Elasticity Based Med-Cloud Recommendation System for Diabetic Prediction in Cloud Computing Environment

Day to day huge medical data have been accumulating for diabetic diseases. The complexity of storing, processing ,analyzing and predicting the data related to diabetics is
not so easy for healthcare professionals .The prediction of accurate results also has the limitation due to scale of data increasing worldwide for patients, symptoms and test
results .In this paper ,it has been tried to considered the diabetic related data storage on cloud and adopt integrated computational algorithms of datamining for better prediction
system to various diabetic types(Type 1, Type 2 and Gestational).Though many computational prediction model and recommended system have been proposed by many researchers ,the proposed model has the novelty of considering the elasticity in data analysis due to frequent data changes of patients due to diabetic test time to time. In this work, Elasticity based Med-Cloud Recommendation System (EMCRS) is proposed for predicting the diabetic disease types and providing recommendations for the patients diagnosed with diabetes. Moreover, elastic resource allocation mechanism is proposed to provide cloud resources an on-demand basis to EMCRS.Various computational algorithms have been used for different proposed to make EMCRS to predict results as compared other existing system. The Adaptively Toggle Genetic Algorithm (ATGA) is applied for elastic resource allocation while increase in the number of data sets. ATGA has taken toggle genetic algorithm that shifts between Roulette Wheel Selection Operator. Hybrid Classification and Clustering Algorithm (HC2A) is used for classifying and clustering the diseased patients as Type 1, Type 2 and Gestational Diabetic patients. Fuzzy C Means clustering based attribute weighting (FCMAW) was used for classifying the diabetic data set. The accuracy of the system tested on Pima Indian Diabetic Dataset (PID) and US Diabetic Dataset (USD) from UCI website which is approximately 98% classification accuracy.