Accelerometer-based body-position sensing for ambulatory electrocardiographic monitoring.

Our objective is to validate the ability of 3 appropriately placed accelerometers to determine body position during ambulatory electrocardiographic (ECG) monitoring and to demonstrate the clinical applicability of this method. During ambulatory (Holter) monitoring, the ability to know a patient's position (lying down, sitting, standing, or changing from one position to another) is important in the evaluation of common symptoms such as dizziness, palpitations, and syncope. Changes in body position are also known to alter the electrical axis of the heart, resulting in artifactual changes in QRS amplitude and ST-segment morphology. We have developed an ambulatory patient-monitoring instrument that, through the use of microfabricated accelerometers, can simultaneously record body-position information and 2 channels of ECG data. The accelerometers measure the effects of gravity and dynamic acceleration, allowing determination of a patient's orientation and movements. The accelerometer and ECG signals are input to a portable recorder and are filtered and digitized. Algorithms were developed to automatically determine body position. Ten healthy volunteers wore the device for 1 hour and followed a protocol of standing, sitting, walking, lying supine, and lying in the left and right lateral decubitus positions. An observer manually recorded times of position changes. Data were recorded and analyzed using software designed with MATLAB. The ability of the accelerometers and computer algorithms to determine body position was analyzed in terms of the sensitivity and specificity for each body position. The sensitivities for sitting, standing, walking, lying supine, lying right, and lying left were 98.8%, 99.2%, 95.5%, 99.1%, 98.9%, and 94.8%, respectively. The specificities were 99.7%, 99.4%, 99.6%, 99.0%, 99.8%, and 99.9%, respectively. The use of microfabricated accelerometers is a clinically feasible method to determine body position and can be applied to future studies correlating body position with ECG or other physiologic data.