e-ISSN : 2455 - 5371

Efficient Mining of Cloud Based Electronic Health Records (EHR) for Clinical Decision Support System

Paper Topic :

Data Mining

Author Name :

Konica Dhingra

Abstract :

In feats to support the growing trend of maintaining huge amount of clinical data in a common storage (Electronic Health Records) and predicting diseases based on the results of mining these data. Community clouds are in general trend where patient’s medical history is stored and based on this training datasets, the diseases of new incoming patient is predicted using Clinical Decision Support System (CDSS). Data mining algorithms provide an efficient way to guess the disease of new incoming patient. The aim of this research focuses on developing parallelized classification algorithms which effectively uses computational benefits of cloud storage to accurately predict the disease. Our algorithm achieves parallelism by dividing the datasets into sub portions and feeding each sub portion into an individual processor (virtual machine) on cloud and then applying algorithms like Decision Tree induction, k-n-n classifier and Naïve Bayesian classifier. Laplacian correction technique is used to remove attributes whose values may be zero .At last, we have used ensemble methods like Random Forests and Bagging to find out the class with majority votes and assign that class to the incoming tuple (Patient’s symptoms) which acts as the output of the CDSS.

Download Article