A body position detection method by fusing heterogeneous information from surface ECG

Determination of body position is a very important issue in biomedical and healthcare areas. The aim of this research is to propose a body position detection method by fusing multiple heterogeneous features from three-lead surface ECG. Our results indicate that the heart axis is more accurate than HRV and PR intervals for posture detection. In addition, for standing and lying classification only, 99.93% training and 66.67% testing accuracy can be achieved for system performance. However, if a subject's identity is known in advance by using ECG biometrics, the performance may be further improved. Overall, ECG is potentially able to combine with other external signals to provide more reliable position detection on homecare systems for prevention of false alarm.