ECG-based detection of body position changes using a Laplacian noise model

Body position changes (BPC), which are often manifested in the ECG as shifts in the electrical axis of the heart, result in ST changes, and thus, may be misclassified as ischemic events during ambulatory monitoring. We have developed a BPC detector by modeling shifts as changes in the Karhunen-Loève transform coefficients of the QRS complex and the ST-T waveform. The noise is assumed to have a Laplacian distribution. A generalized likelihood ratio test has been chosen as the strategy to detect BPCs. Two different databases have been used to assess detection performance. The obtained results were 93%/99% in terms of sensitivity/positive predictivity value (S/+PV) and a false alarm rate of 2 events/hour. The results clearly outperform current techniques (S/+PV: 85%/99%) based on the Gaussian noise assumption.