Project outline
This project is a strategic collaboration with the Yorkshire Ambulance Service (YAS) aimed at enhancing the speed and accuracy of emergency call prioritization. The core objective is to develop and validate advanced deep learning models capable of analyzing the acoustic properties of emergency calls in real-time.
A significant challenge in emergency dispatch is the rapid identification of life-threatening conditions, such as cardiac arrest, where every second is critical. This research focuses on detecting subtle, non-linguistic acoustic cues—including labored breathing, vocal strain, and indicators of extreme physiological distress—that may be difficult for human call handlers to identify consistently under high-pressure conditions.
By automating the extraction of these features, the project seeks to provide a robust decision-support tool for dispatch centers. The ultimate goal is to predict Category 1 emergencies with higher precision, reducing response times and integrating AI-driven insights into the frontline of emergency care to improve patient outcomes and save lives.
Project Partners
This project is hosted at the University of Sheffield and is supported by Yorkshire Ambulance Service.
Student
Shumin Que