Reflections on AI- and sensor-integrated versus face-to-face CPR reskilling for chest compression eminence

Dear Editor,

We read with great interest the randomized trial by Ma et al. ¹ evaluating the immediate learning outcomes of an artificial intelligence (AI) and sensor-integrated cardiopulmonary resuscitation (CPR) retraining system compared with conventional face-to-face instruction among healthcare professionals. The study provides an important contribution to technology-enhanced resuscitation education, particularly in its structured methodology, use of objective performance metrics, and integration of AI-driven feedback systems.

Several limitations need consideration. One is the exclusive focus on immediate post-training outcomes, which bounds insight into long-term skill retention. CPR competency tends to deteriorate over time without reinforcement, and evaluating only immediate performance may overestimate the intervention’s sustained effectiveness. ²

Moreover, there is a lack of blinding, which, although challenging in educational interventions, may introduce performance bias. ³

Furthermore, the study population’s inadequate generalizability. The trial included healthcare professionals from a single hospital or region, which may not reflect the diversity of learners in other healthcare settings or geographic areas. Factors such as prior CPR experience, cultural differences in training approaches, and variations in local resuscitation protocols can influence learning outcomes. Consequently, the findings may not be fully applicable to broader populations, including laypersons or first responders in community settings, where bystander CPR is critical for survival. ²

Additionally, there is an imbalance in feedback mechanisms between the study groups. The AI-integrated group received real-time, objective feedback, whereas the traditional group relied on subjective instructor assessment. Recent evidence indicates that real-time feedback significantly improves CPR performance, making it difficult to isolate the effect of the training modality alone. ⁴

In conclusion, while the study by Ma et al. provides important insights into AI-supported CPR training, future studies incorporating longitudinal follow-up, diverse populations, and balanced feedback mechanisms will be essential to clarify the independent impact of AI-based training systems on CPR performance.