FURI | Fall 2025

Transcribing EMS Calls Using Deep Learning for Clinical Decision Support

Health icon, disabled. A red heart with a cardiac rhythm running through it.

This project fine-tunes the Whisper speech-to-text model to transcribe emergency medical services (EMS) calls accurately. EMS calls often contain background noise, interruptions, and medical jargon that hinder reliable transcription. By training Whisper on EMS-specific audio, the research improves transcription quality and preserves clinical details. These transcripts form the foundation for automated information extraction systems that predict patient needs and support rapid triage. The outcome enables a seamless pipeline from audio to structured data, advancing efficiency in emergency healthcare.