EMSDialog: Synthetic Multi-person Emergency Medical Service Dialogue Generation from Electronic Patient Care Reports via Multi-LLM Agents

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ge, Xueren, Murtaza, Sahil, Cortez, Anthony, Alemzadeh, Homa
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911607774248960
author Ge, Xueren
Murtaza, Sahil
Cortez, Anthony
Alemzadeh, Homa
author_facet Ge, Xueren
Murtaza, Sahil
Cortez, Anthony
Alemzadeh, Homa
contents Conversational diagnosis prediction requires models to track evolving evidence in streaming clinical conversations and decide when to commit to a diagnosis. Existing medical dialogue corpora are largely dyadic or lack the multi-party workflow and annotations needed for this setting. We introduce an ePCR-grounded, topic-flow-based multi-agent generation pipeline that iteratively plans, generates, and self-refines dialogues with rule-based factual and topic flow checks. The pipeline yields EMSDialog, a dataset of 4,414 synthetic multi-speaker EMS conversations based on a real-world ePCR dataset, annotated with 43 diagnoses, speaker roles, and turn-level topics. Human and LLM evaluations confirm high quality and realism of EMSDialog using both utterance- and conversation-level metrics. Results show that EMSDialog-augmented training improves accuracy, timeliness, and stability of EMS conversational diagnosis prediction. Our datasets and code are publicly available at https://uva-dsa.github.io/EMSDialog
format Preprint
id arxiv_https___arxiv_org_abs_2604_07549
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle EMSDialog: Synthetic Multi-person Emergency Medical Service Dialogue Generation from Electronic Patient Care Reports via Multi-LLM Agents
Ge, Xueren
Murtaza, Sahil
Cortez, Anthony
Alemzadeh, Homa
Computation and Language
Artificial Intelligence
Conversational diagnosis prediction requires models to track evolving evidence in streaming clinical conversations and decide when to commit to a diagnosis. Existing medical dialogue corpora are largely dyadic or lack the multi-party workflow and annotations needed for this setting. We introduce an ePCR-grounded, topic-flow-based multi-agent generation pipeline that iteratively plans, generates, and self-refines dialogues with rule-based factual and topic flow checks. The pipeline yields EMSDialog, a dataset of 4,414 synthetic multi-speaker EMS conversations based on a real-world ePCR dataset, annotated with 43 diagnoses, speaker roles, and turn-level topics. Human and LLM evaluations confirm high quality and realism of EMSDialog using both utterance- and conversation-level metrics. Results show that EMSDialog-augmented training improves accuracy, timeliness, and stability of EMS conversational diagnosis prediction. Our datasets and code are publicly available at https://uva-dsa.github.io/EMSDialog
title EMSDialog: Synthetic Multi-person Emergency Medical Service Dialogue Generation from Electronic Patient Care Reports via Multi-LLM Agents
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2604.07549