EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

MEGA-RAG: a retrieval-augmented generation framework with multi-evidence guided answer refinement for mitigating hallucinations of LLMs in public health

Shan Xu, Zhaokun Yan, Chengxiao Dai, Fan Wu

Frontiers in Public Health · 08.10.2025 · 13

DOI: 10.3389/fpubh.2025.1635381 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Public Health, online erschienen am 08.10.2025. Verfasst von Shan Xu und 3 weiteren Autor:innen. Beteiligte Einrichtung: Shanghai Institute of Infectious Disease and Biosecurity, School of Public Health, Fudan University, Shanghai, China. Hauptthemen laut MEDLINE: Hallucinations, Public Health, Language, Information Storage and Retrieval. Die Arbeit wurde bisher 55-mal zitiert (Crossref) und nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

The increasing adoption of large language models (LLMs) in public health has raised significant concerns about hallucinations-factually inaccurate or misleading outputs that can compromise clinical communication and policy decisions.

Methods

We propose a retrieval-augmented generation framework with multi-evidence guided answer refinement (MEGA-RAG), specifically designed to mitigate hallucinations in public health applications. The framework integrates multi-source evidence retrieval (dense retrieval via FAISS, keyword-based retrieval via BM25, and biomedical knowledge graphs), employs a cross-encoder reranker to ensure semantic relevance, and incorporates a discrepancy-aware refinement module to further enhance factual accuracy.

Results

Experimental evaluation demonstrates that MEGA-RAG outperforms four baseline models [PubMedBERT, PubMedGPT, standalone LLM, and LLM with standard retrieval-augmented generation (RAG)], achieving a reduction in hallucination rates by over 40%. It also achieves the highest accuracy (0.7913), precision (0.7541), recall (0.8304), and F1 score (0.7904).

Discussion

These findings confirm that MEGA-RAG is highly effective in generating factually reliable and medically accurate responses, thereby enhancing the credibility of AI-generated health information for applications in health education, clinical communication, and evidence-based policy development.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

HallucinationsPublic HealthLanguageInformation Storage and RetrievalHumans

Stichwörter der Autor:innen

AI in healthcareLLM hallucinationsRAGmedical question answeringpublic health

Autor:innen und Institutionen

  • Shanghai Institute of Infectious Disease and Biosecurity, School of Public Health, Fudan University, Shanghai, China.
  • China Academy of Information and Communications Technology, Beijing, China.
  • Faculty of Engineering, University of Sydney, Sydney, NSW, Australia.

Zitieren

APA

Xu, S., Yan, Z., Dai, C., & Wu, F. (2025). MEGA-RAG: a retrieval-augmented generation framework with multi-evidence guided answer refinement for mitigating hallucinations of LLMs in public health. Frontiers in Public Health, 13. https://doi.org/10.3389/fpubh.2025.1635381

Vancouver

Xu S, Yan Z, Dai C, Wu F. MEGA-RAG: a retrieval-augmented generation framework with multi-evidence guided answer refinement for mitigating hallucinations of LLMs in public health. Frontiers in Public Health. 2025;13. doi:10.3389/fpubh.2025.1635381

BibTeX
@article{Xu2025,
  author = {Xu, Shan and Yan, Zhaokun and Dai, Chengxiao and Wu, Fan},
  title = {MEGA-RAG: a retrieval-augmented generation framework with multi-evidence guided answer refinement for mitigating hallucinations of LLMs in public health},
  journal = {Frontiers in Public Health},
  year = {2025},
  volume = {13},
  doi = {10.3389/fpubh.2025.1635381}
}
Kostenlos starten →