| description abstract | Abstract. Pediatric cardiopulmonary resuscitation (CPR) quality remains inconsistent, with low adherence to guideline-recommended compression rate and depth. Augmented reality cardiopulmonary resuscitation (AR-CPR) is an augmented reality feedback system designed to improve CPR performance using real-time, in-view coaching via smart glasses. To validate the accuracy and precision of the AR-CPR system in measuring chest compression rate and depth across clinically relevant ranges, we tested AR-CPR using a programable oscillation platform at five rates (100–140 compressions per minute (CPM)) and four depths (4.0–5.5 cm), and with the Stryker LUCAS3 device at 102 CPM and 5.3 cm. A total of 473 compressions (slide test) and 559 compressions (LUCAS3 test) were analyzed. Statistical methods included intraclass correlation coefficients (ICC[2,1]), paired t-tests, Bland–Altman analysis, root-mean-square error (RMSE), kernel density estimation for error distribution, and group error modeling to estimate clinical thresholds (±2.5 CPM, ±0.5 cm). Linearity was assessed via linear regression. The AR-CPR system demonstrated high accuracy and reliability in 473 simulated and 559 mechanical compressions. Mean biases were minimal for rate (−0.48, −0.44 CPM) and depth (+0.39, +0.59 cm), with excellent ICCs (0.997 rate, 0.944 depth). Errors were normally distributed, with <7% exceeding clinically relevant thresholds. R2 values (0.994 rate, 0.903 depth) confirmed strong linear agreement with reference values. AR-CPR reliably measured compression rate and depth with high accuracy and precision across variable and fixed testing conditions. Its portability, real-time feedback, and robust signal processing support its potential for improving pediatric resuscitation training and clinical performance. | |
