MedPAO: A Protocol-Driven Agent for Structuring Medical Reports

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Vaidya, Shrish Shrinath, Palani, Gowthamaan, Ramesh, Sidharth, Balasubramanian, Velmurugan, Selvam, Minmini, Srinivasaraja, Gokulraja, Krishnamurthi, Ganapathy
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908577496563712
author Vaidya, Shrish Shrinath
Palani, Gowthamaan
Ramesh, Sidharth
Balasubramanian, Velmurugan
Selvam, Minmini
Srinivasaraja, Gokulraja
Krishnamurthi, Ganapathy
author_facet Vaidya, Shrish Shrinath
Palani, Gowthamaan
Ramesh, Sidharth
Balasubramanian, Velmurugan
Selvam, Minmini
Srinivasaraja, Gokulraja
Krishnamurthi, Ganapathy
contents The deployment of Large Language Models (LLMs) for structuring clinical data is critically hindered by their tendency to hallucinate facts and their inability to follow domain-specific rules. To address this, we introduce MedPAO, a novel agentic framework that ensures accuracy and verifiable reasoning by grounding its operation in established clinical protocols such as the ABCDEF protocol for CXR analysis. MedPAO decomposes the report structuring task into a transparent process managed by a Plan-Act-Observe (PAO) loop and specialized tools. This protocol-driven method provides a verifiable alternative to opaque, monolithic models. The efficacy of our approach is demonstrated through rigorous evaluation: MedPAO achieves an F1-score of 0.96 on the critical sub-task of concept categorization. Notably, expert radiologists and clinicians rated the final structured outputs with an average score of 4.52 out of 5, indicating a level of reliability that surpasses baseline approaches relying solely on LLM-based foundation models. The code is available at: https://github.com/MiRL-IITM/medpao-agent
format Preprint
id arxiv_https___arxiv_org_abs_2510_04623
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MedPAO: A Protocol-Driven Agent for Structuring Medical Reports
Vaidya, Shrish Shrinath
Palani, Gowthamaan
Ramesh, Sidharth
Balasubramanian, Velmurugan
Selvam, Minmini
Srinivasaraja, Gokulraja
Krishnamurthi, Ganapathy
Artificial Intelligence
The deployment of Large Language Models (LLMs) for structuring clinical data is critically hindered by their tendency to hallucinate facts and their inability to follow domain-specific rules. To address this, we introduce MedPAO, a novel agentic framework that ensures accuracy and verifiable reasoning by grounding its operation in established clinical protocols such as the ABCDEF protocol for CXR analysis. MedPAO decomposes the report structuring task into a transparent process managed by a Plan-Act-Observe (PAO) loop and specialized tools. This protocol-driven method provides a verifiable alternative to opaque, monolithic models. The efficacy of our approach is demonstrated through rigorous evaluation: MedPAO achieves an F1-score of 0.96 on the critical sub-task of concept categorization. Notably, expert radiologists and clinicians rated the final structured outputs with an average score of 4.52 out of 5, indicating a level of reliability that surpasses baseline approaches relying solely on LLM-based foundation models. The code is available at: https://github.com/MiRL-IITM/medpao-agent
title MedPAO: A Protocol-Driven Agent for Structuring Medical Reports
topic Artificial Intelligence
url https://arxiv.org/abs/2510.04623