Abstract Body: Introduction
High-quality CPR with minimal interruptions is a key component of successful resuscitation from out-of-hospital cardiac arrest (OHCA), but current approaches require pauses in CPR to assess for a pulse. Finger pulse oximeters utilize photoplethysmograms (PPG) and can detect pulsatility by identifying peaks corresponding to individual cardiac cycles. The use of PPG for identifying spontaneous circulation during ongoing CPR has not been previously explored.

Aim
We sought to evaluate changes in the PPG signal during CPR and compare signal characteristics among those who achieved return of spontaneous circulation (ROSC) during subsequent pulse checks versus those who did not.

Methods
We performed a retrospective study of adult out of hospital cardiac arrests treated by EMS from 2017-2023 in Seattle, WA. Each case contained electronic defibrillator recordings including ECG, transthoracic impedance, capnography, and PPG. Cases were time-stamped for events based on review of the audio channel, including start and stop times for CPR, return of spontaneous circulation (ROSC), and presence of a palpable pulse. PPG segments in the 4 seconds preceding manual pulse checks were extracted during CPR and categorized based on the presence or absence of ROSC during the ensuing CPR pause for manual pulse check. A model was created using features of the Fourier transform of the PPG signal. Initial features included mean, median, number of peaks, max frequency, and magnitude of the max frequency.

Results
Using 1557 eligible patients, 1569 PPG segments were isolated during CPR. 43 (3%) preceded ROSC and 1526 (97%) did not. The scalogram of the segments in Figure 1 demonstrates detectable differences in frequency magnitudes. The multivariate ROSC detection model applied to segments without CPR had an AUC 0.63969 (Figures 2) with p = 0.00001 using leave one out cross validation.

Conclusion
PPG signals measured via finger pulse oximeters demonstrate distinct and detectable characteristics without CPR among patients who achieve ROSC on subsequent manual pulse checks compared with those who do not. These findings suggest the potential for automated pulse detection using PPG during resuscitation efforts, eliminating the need for interruptions to perform manual pulse checks. Such advancements could enhance CPR delivery and improve resuscitation outcomes. Future models will incorporate features from the scalograms to improve prediction.