Feedback derived from 3D Accelerometers used for Autonomous Correction of Chest Compression Movement in Cardiopulmonary Resuscitation Training

This paper reports on an exploration whether feedback, derived from the comparison of a programmed baseline and actual chest compressions during cardiopulmonary resuscitation (CPR), will help trainees to improve their chest compression movement without the need for personal training by a coach. A prototype is constructed that is used to investigate if it is possible to reliably detect chest compression movement and derive feedback from the comparison of that movement to a programmed baseline. Several methods to present the resulting feedback are implemented as well. An experiment is set up to test these feedback methods and to see how well the derived feedback allows the trainees to improve their skills. The results show that reliable feedback can be derived from the 3D accelerometer of a SunSPOT device attached to the back of the hand of the person performing chest compressions. It is also shown that trainees markedly improve their chest compression-skills after the feedback is presented to them. The method of feedback that uses audio is preferred by trainees, but measurements show a combination of audio, text and LED-based feedback leads to the greatest improvement of chest compression skills. Keyword