Stand-Alone Heartbeat Detection in Multidimensional Mechanocardiograms

We describe a home health monitoring solution with cardiac beat-to-beat detection using accelerometer and gyroscope signal fusion. The proposed method measures both the precordial translational and rotational motions of the chest using miniaturized inertial sensors. The algorithm removes motion artefacts, selects the best axis from multi-axial accelerometric and gyroscopic signals and detects the location of beats using two detection principles based on the signal envelope and signal morphology. We consider the beat-to-beat detection accuracy, estimate the heart rate and compare the detection performance between the sensor modalities in two study groups: i) healthy subjects and ii) heart disease patients. The average sensitivity and precision of the beat detection were 99.9% and 99.6% for the healthy subjects and 96.1% and 95.6% for the heart disease patients, respectively. Although high-accuracy beat detection was achieved for the heart disease patients, location matching in these patients was found to be less accurate compared to that of the healthy subjects. The average root mean square error (RMSE) between the mechanical and ECG interbeat intervals was 5.6 ms for the healthy patients; this error increased approximately 10-fold for the heart disease patients. Similarly, the RMSE for the averaged heart rate estimation showed about a 10-fold difference at 1.05 beats per minute for the heart disease patients. The used sensor modalities are found in many electronic devices, such as smartphones and wearable technologies and this method provides a step towards ubiquitous cardiac monitoring.