Automatic Outlier Detection And Data Consistency Maintenance Technique For Iot Healthcare Applications

Abstract

In a typical health-care system, various sensor nodes are employed for monitoring different vital parameters of a patient. In Internet of Things (IoT) based healthcare systems, data correctness is essential since clinical decisions often depend on real-time physiological data collected from sensors and wearable devices. In this paper, an automatic outlier detection and data consistency maintenance (AOD-DCM) technique for IoT-WSN is proposed. It consists of data outlier detection and data inconsistency checking phases. For data outlier detection, the Principal Component Analysis (PCA) algorithm is applied.  For data inconsistency checking from existing data readings, the Deep Auto Encoder (DAE) model is designed. To fine tune the parameters of DAE, the bio inspired Brown Bear Optimization (BBO) algorithm is applied. Experimental results show that DAE-BBO technique achieves higher detection accuracy and correctness of data, when compared to the existing outlier detection techniques.