Abnormality detection on vital parameters using modified weighted average method in cloud

Abstract

Personal Health Record (PHR) is an online service model that holds patient's vital parameter data from the sensors worn by the patient. It allows patients to easily share their health information from any location with doctors. An android smartphone fetches vital signs of the patient from the sensors configured with the smartphone. The various vital signs are grouped together on a time to time basis as a record (i.e., PHR) and is then uploaded to a cloud storage through the smartphone. The PHRs may contain abnormal vital signs. An algorithm running on the private cloud constantly monitors the vital signs streamed from the smartphone to detect any abnormality in the patient's health condition. Existing algorithms omits the patient's health history while detecting abnormality. To overcome this existing issue, a cloud based health monitoring system has been developed using a Modified Weighted Average Method (MWAM) along with Naive Bayesian classifier is proposed. The proposed Cloud based Remote Monitoring System(CRMS) classifies whether the patient's health condition is normal or abnormal by giving equal preference to patient's health history and current vital signs and also predicts the degree of abnormality on a scale of 1 to 3 (low, medium and high). An android application has been developed for the patients that receives live data from the WBAN sensors and uploads to the cloud. Abnormality detection algorithm is deployed on the cloud setup that continuously monitors the patient's vital parameters and alerts the doctor and patient's caretakers when the patient's health is abnormal. From the experimental results, it is found that the MWAM boosts the performance of the Naive Bayesian algorithm in alert generation.