Guardado en:
Detalles Bibliográficos
Autores principales: Wang, Yinuo, Wang, Baiyang, Mercer, Robert E., Rudzicz, Frank, Roy, Sudipta Singha, Ren, Pengjie, Chen, Zhumin, Wang, Xindi
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2506.03659
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912682211278848
author Wang, Yinuo
Wang, Baiyang
Mercer, Robert E.
Rudzicz, Frank
Roy, Sudipta Singha
Ren, Pengjie
Chen, Zhumin
Wang, Xindi
author_facet Wang, Yinuo
Wang, Baiyang
Mercer, Robert E.
Rudzicz, Frank
Roy, Sudipta Singha
Ren, Pengjie
Chen, Zhumin
Wang, Xindi
contents Trustworthiness in healthcare question-answering (QA) systems is important for ensuring patient safety, clinical effectiveness, and user confidence. As large language models (LLMs) become increasingly integrated into medical settings, the reliability of their responses directly influences clinical decision-making and patient outcomes. However, achieving comprehensive trustworthiness in medical QA poses significant challenges due to the inherent complexity of healthcare data, the critical nature of clinical scenarios, and the multifaceted dimensions of trustworthy AI. In this survey, we systematically examine six key dimensions of trustworthiness in medical QA, i.e., Factuality, Robustness, Fairness, Safety, Explainability, and Calibration. We review how each dimension is evaluated in existing LLM-based medical QA systems. We compile and compare major benchmarks designed to assess these dimensions and analyze evaluation-guided techniques that drive model improvements, such as retrieval-augmented grounding, adversarial fine-tuning, and safety alignment. Finally, we identify open challenges-such as scalable expert evaluation, integrated multi-dimensional metrics, and real-world deployment studies-and propose future research directions to advance the safe, reliable, and transparent deployment of LLM-powered medical QA.
format Preprint
id arxiv_https___arxiv_org_abs_2506_03659
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Trustworthy Medical Question Answering: An Evaluation-Centric Survey
Wang, Yinuo
Wang, Baiyang
Mercer, Robert E.
Rudzicz, Frank
Roy, Sudipta Singha
Ren, Pengjie
Chen, Zhumin
Wang, Xindi
Computation and Language
Trustworthiness in healthcare question-answering (QA) systems is important for ensuring patient safety, clinical effectiveness, and user confidence. As large language models (LLMs) become increasingly integrated into medical settings, the reliability of their responses directly influences clinical decision-making and patient outcomes. However, achieving comprehensive trustworthiness in medical QA poses significant challenges due to the inherent complexity of healthcare data, the critical nature of clinical scenarios, and the multifaceted dimensions of trustworthy AI. In this survey, we systematically examine six key dimensions of trustworthiness in medical QA, i.e., Factuality, Robustness, Fairness, Safety, Explainability, and Calibration. We review how each dimension is evaluated in existing LLM-based medical QA systems. We compile and compare major benchmarks designed to assess these dimensions and analyze evaluation-guided techniques that drive model improvements, such as retrieval-augmented grounding, adversarial fine-tuning, and safety alignment. Finally, we identify open challenges-such as scalable expert evaluation, integrated multi-dimensional metrics, and real-world deployment studies-and propose future research directions to advance the safe, reliable, and transparent deployment of LLM-powered medical QA.
title Trustworthy Medical Question Answering: An Evaluation-Centric Survey
topic Computation and Language
url https://arxiv.org/abs/2506.03659