A data fusion algorithm for clinically relevant anomaly detection in remote health monitoring

The adoption of the Internet of Things (IoT) technologies in healthcare lead to the Healthcare 4.0 paradigm. In this paradigm, Remote Health Monitoring (RHM) applications emerge to provide continuous monitoring of patient's health conditions. But RHM applications traditionally present high rates of false alarms. This disturbance is caused by many factors, from the high sensibility of the equipments to real variations in the monitored vital signs not related to emergencies of health degradation. Hence, this work proposes a system for detection and evaluation of medical emergencies, using Wireless Body Sensor Network as its network infrastructure, able to distinguish real emergencies from other cases by considering a risk estimation from each sampled data. Experiments showed that the proposed system can reach an average accuracy rate of 93.0% and detection rate of 87.2%, and an energy consumption profile feasible to WBSN contexts.