Cardiopulmonary Resuscitation Support: Comparison of Wrist-, Chest-, and Ear-Worn Devices and Estimation Algorithms

Abstract

Chest compressions (CC) are the most important means of treating cardiac arrest but are challenging to perform. Real-Time feedback of depth and rate can improve CC quality. For the first time, we compared three wearable positions and six depth, and rate estimation algorithms under the same conditions. We share their optimal tuning parameters. Our evaluation of earables results in a new prime candidate for high-quality cardiopulmonary resuscitation (CPR) feedback. For depth estimation on chest, wrist, and ear the median absolute deviation (MAD) was 3.4 mm, 4.5 mm, and 5.9 mm, respectively (target depth range: 50-60 mm). Though not necessary for effective CC, fusing sensor locations reduces the depth MAD further to 3.2 mm. CC rate was estimated at less than 1.6 compressions per minute (cpm) MAD in all configurations. Hence, all wearables and algorithms give precise input for live-saving CPR.