ReXTrust: A Model for Fine-Grained Hallucination Detection in AI-Generated Radiology Reports

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hardy, Romain, Kim, Sung Eun, Ro, Du Hyun, Rajpurkar, Pranav
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909470672551936
author Hardy, Romain
Kim, Sung Eun
Ro, Du Hyun
Rajpurkar, Pranav
author_facet Hardy, Romain
Kim, Sung Eun
Ro, Du Hyun
Rajpurkar, Pranav
contents The increasing adoption of AI-generated radiology reports necessitates robust methods for detecting hallucinations--false or unfounded statements that could impact patient care. We present ReXTrust, a novel framework for fine-grained hallucination detection in AI-generated radiology reports. Our approach leverages sequences of hidden states from large vision-language models to produce finding-level hallucination risk scores. We evaluate ReXTrust on a subset of the MIMIC-CXR dataset and demonstrate superior performance compared to existing approaches, achieving an AUROC of 0.8751 across all findings and 0.8963 on clinically significant findings. Our results show that white-box approaches leveraging model hidden states can provide reliable hallucination detection for medical AI systems, potentially improving the safety and reliability of automated radiology reporting.
format Preprint
id arxiv_https___arxiv_org_abs_2412_15264
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle ReXTrust: A Model for Fine-Grained Hallucination Detection in AI-Generated Radiology Reports
Hardy, Romain
Kim, Sung Eun
Ro, Du Hyun
Rajpurkar, Pranav
Computation and Language
Artificial Intelligence
The increasing adoption of AI-generated radiology reports necessitates robust methods for detecting hallucinations--false or unfounded statements that could impact patient care. We present ReXTrust, a novel framework for fine-grained hallucination detection in AI-generated radiology reports. Our approach leverages sequences of hidden states from large vision-language models to produce finding-level hallucination risk scores. We evaluate ReXTrust on a subset of the MIMIC-CXR dataset and demonstrate superior performance compared to existing approaches, achieving an AUROC of 0.8751 across all findings and 0.8963 on clinically significant findings. Our results show that white-box approaches leveraging model hidden states can provide reliable hallucination detection for medical AI systems, potentially improving the safety and reliability of automated radiology reporting.
title ReXTrust: A Model for Fine-Grained Hallucination Detection in AI-Generated Radiology Reports
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2412.15264