Body movement activity recognition for ambulatory cardiac monitoring.

Wearable electrocardiogram (W-ECG) recorders are increasingly in use by people suffering from cardiac abnormalities who also choose to lead an active lifestyle. The challenge presently is that the ECG signal is influenced by motion artifacts induced by body movement activity (BMA) of the wearer. The usual practice is to develop effective filtering algorithms which will eliminate artifacts. Instead, our goal is to detect the motion artifacts and classify the type of BMA from the ECG signal itself. We have recorded the ECG signals during specified BMAs, e.g., sitting still, walking, movements of arms and climbing stairs, etc. with a single-lead system. The collected ECG signal during BMA is presumed to be an additive mix of signals due to cardiac activities, motion artifacts and sensor noise. A particular class of BMA is characterized by applying eigen decomposition on the corresponding ECG data. The classification accuracies range from 70% to 98% for various class combinations of BMAs depending on their uniqueness based on this technique. The above classification is also useful for analysis of P and T waves in the presence of BMA.