Can Artificial Intelligence Help Emergency Responders Save Children?
Boston University researchers are running hundreds of high-fidelity pediatric emergency simulations (with computer-controlled mannequins) and recording how EMS teams respond. In the near term, the work is about understanding whether real-time video calls with physicians improve outcomes in the field—especially when responders need to choose and execute time-critical interventions.
The longer-term goal is to turn those recordings into training data for an AI “copilot” responders could carry on a phone or laptop. The motivating constraint is that severely ill children are rare in prehospital care, but high-stakes when they happen: dosing is different, procedures are stressful, and teams may have limited pediatric experience. The study is designed to capture the granular “time to critical intervention” details that are hard to learn from retrospective charts.
If the approach works, the same interface could eventually help with basics like CPR quality (compression depth/rate, reminders to ventilate) or medication checks in noisy, high-pressure scenes. The team is planning 500+ recorded observations across multiple EMS agencies in Massachusetts and eight other states, supported by a multi-year NIH grant.