From stress to success: using physiological data to predict cardiopulmonary resuscitation simulation performance

TL;DR

This study uses physiological data to predict performance in CPR simulations, showing that stress responses and autonomic markers are linked to success.

Contribution

The study introduces a novel use of physiological data to predict CPR performance and identifies nonlinear relationships between autonomic markers and success.

Findings

Physiological stress markers like HRV, HR, and EDA increased during CPR simulations.

Sympathetic activation in team leaders predicted overall CPR performance.

A nonlinear relationship between HRV-derived SD1 and performance was identified.

Abstract

Managing stress is critical in emergency medicine, where cardiopulmonary resuscitation (CPR) rely on team dynamics. Although subjective and physiological markers assess stress, few studies have examined their combined effects during CPR simulations. The influence of team role (leader vs. member) and whether physiological data can predict performance also remain underexplored. This study addresses these gaps.

Thirty emergency residents attending the School of Emergency Medicine of the University of Padua (Italy) were recruited with previous certification in Advance Cardiac Life Support (ACLS) and randomly paired, each assigned to one of two roles: team leader (TL) or team member (TM). Randomization also considered baseline stress level (PSS-10). Each pair was then assigned to cardiac arrest with a shockable or non-shockable rhythm, including 2 min of uninterrupted chest compressions, following American Heart Association (AHA) guidelines. The data collected included CPR performance metrics (compression depth, rate, recoil). Physiological data were collected before, during, and after CPR using Empatica E4 and eSense, Heart rate (HR), Heart Rate Variability (HRV), Electrodermal Activity (EDA), and Skin conductance response (SCR).

Participants reported moderate baseline stress (PSS-10, VAS stress/anxiety). Baseline physiological measures were within normative ranges. ANOVA revealed a significant effect of group condition for HRV (p < 0.05); HR significantly increased from baseline to CPR (p < 0.001) and decreased post-CPR (p < 0.001). EDA increased from baseline to both CPR and post-CPR (p < 0.001). No significant differences were found between team roles at exception for HRV. Binomial logistic regression models using sympathetic data did predict CPR performance (TM EDA Pre, TL EDA Pre and TL SCR pre simulation, R2 = 0.39, AIC = 19.804, p < 0.05, accuracy = 0.8667). Furthermore, a nonlinear regression using HRV-derived SD1 predicted performance (R2 = 0.56; coefficient a, p < 0.01; coefficient b, p < 0.01).

This study shows that simulated CPR scenarios trigger psychophysiological stress responses. Increased HRV, HR, and EDA indicate a challenge-type reaction, despite stable subjective ratings across team roles, suggesting a shared load, with TL sympathetic activation as a possible mediator of global team activation. Notably, a nonlinear link between SD1 and performance emerged, indicating autonomic flexibility relevance.

Created in BioRender, Queirolo, L. (2026), https://BioRender.com/850077a.
Infographic summarizes a study on using physiological data to predict CPR simulation performance, detailing background, methods, and results. Key findings indicate increased heart rate, heart rate variability, and electrodermal activity during simulated CPR, with sympathetic and parasympathetic markers predicting performance. Quadratic regression graph visualizes data relationship. Conclusion highlights psychophysiological stress response and team activation influencing performance.

Figures