Project Details

Harnessing Artificial Intelligence to Augment Team Performance in Neonatal Resuscitation

Eli Cahan, MD, MSc

Summary

BACKGROUND: Neonatal resuscitation team performance varies widely, with studies reporting adherence to resuscitation guidelines ranging from 15% to 55%, contributing in part to the estimated 814,000 neonatal deaths annually from potentially preventable perinatal events. Simulation-based training for quality improvement has been shown to significantly augment team performance and clinical outcomes by enhancing communication and minimizing critical errors during resuscitation. GAP: Traditional models of simulation-based research involve using trained observers to manually code team behaviors on retrospective video review. This analog methodology is time consuming, poorly scalable, vulnerable to observer bias, and analytically limited. HYPOTHESIS: We hypothesize firstly that Artificial intelligence (AI)-informed models can reliably and efficiently classify communication patterns used during simulated neonatal resuscitation (SNR). We hypothesize secondly that AI-informed models can accurately predict high quality versus low quality performance during SNR based on observed communication patterns. METHODS: Videos from an archive maintained by Stanford’s Center for Advanced Perinatal Education will be used to train AI models to interpret audio and visual communication patterns. Primary outcomes will include (a) team performance and (b) time to achieve performance markers. Team performance will be defined using the Neonatal Resuscitation Program Evaluation tool, a validated scale used to grade neonatal resuscitation quality. RESULTS: Pending. IMPACT:AI-driven analysis can be used to discriminate team performance during SNR and identify discrete communication patterns that facilitate adherence to resuscitation guidelines. These findings can inform quality improvement and implementation of best practices capable of reducing perinatal morbidity and mortality in a rapid and scalable manner. Website Link: https://med.stanford.edu/cape.html