Anomaly Detection for Streaming Data from Wearable Sensor Network

Abstract

Wearable health device refers to the using of wearable sensors to collect and monitor biological and physiological parameters of human motion streaming data. Generally it contains heart rate, pulse rate, respiratory rate, body temperature, heat consumption, blood pressure, blood sugar, blood oxygen, hormones and BMI index, body fat content, and etc. It helps users to manage their physiological activities. The objective of this research is to develop an anomaly detection algorithm for data collected from medical wireless sensors. We fist introduce our framework, then focus on representing historical anomalies and the matching algorithms to detect potential anomalies. The experimental results on real patient datasets show that the proposed approach can efficiently detect patients' anomalies and sense fault with high accuracy meanwhile keeping reasonable alarm precision and recall ratios.