Fine-Tuning, Retrieval-Augmented Generation, and Hybrid Large Language Models for Postoperative Decision Support: Comparative Analysis
Journal of Medical Internet Research · 2026-07-14 · 28 · e90692-e90692
DOI: 10.2196/90692 ↗At a glance
Vergleichsstudie in Journal of Medical Internet Research, published online on 2026-07-14. Written by Srinivasagam Prabha and 9 co-authors. Institution: Division of Plastic Surgery, Mayo Clinic in Florida, 4500 San Pablo Road South, Jacksonville, FL, 32224, United States, 1 904-953-2000. Major MEDLINE topics: Postoperative Care, Decision Support Systems, Clinical. 72 references. The full text is freely available in PubMed Central.
Abstract (original language)
Background
Large language models (LLMs) show growing potential for decision support. However, integrating domain-specific medical knowledge while maintaining accuracy, safety, and interpretability remains challenging for postoperative discharge instructions and patient education. Fine-tuning, retrieval-augmented generation (RAG), and hybrid fine-tuning+RAG approaches are prominent strategies for knowledge integration, but their comparative performance in postoperative care has not been systematically evaluated.
Objective
We aimed to compare the performance, reliability, and safety characteristics of baseline, fine-tuning, RAG, and hybrid fine-tuning+RAG LLM configurations for postoperative decision support.
Methods
We conducted a comparative evaluation of 4 LLM configurations using Google Gemini 2.5 Flash. A total of 600 postoperative question and answer pairs were used for model adaptation and validation, while 150 queries were reserved for final evaluation. Queries included routine postoperative questions, emergency escalation scenarios, and deliberately out-of-scope prompts. Outputs were independently assessed by 3 blinded clinical experts for clinical medical accuracy, safety or refusal accuracy, completeness, and relevance. Automated metrics evaluated readability, faithfulness, and hallucination propensity.
Results
All knowledge-enhanced models significantly outperformed baseline in overall accuracy (baseline 68% vs fine-tuning 92.7%, RAG 91.3%, fine-tuning+RAG 97.3%; P
Subjects
MeSH terms (MEDLINE)
Author keywords
Authors and institutions
- Division of Plastic Surgery, Mayo Clinic in Florida, 4500 San Pablo Road South, Jacksonville, FL, 32224, United States, 1 904-953-2000.
- Division of Clinical Trials and Biostatistics, Mayo Clinic in Florida, Jacksonville, FL, United States.
- Department of Radiology AI IT, Mayo Clinic, Rochester, MN, United States.
- Department of Surgery, Mayo Clinic in Florida, Jacksonville, FL, United States.
- Department of Artificial Intelligence and Informatics, Mayo Clinic in Florida, Jacksonville, FL, United States.
- Center for Digital Health, Mayo Clinic, Rochester, MN, United States.
ORCID profiles: Srinivasagam Prabha, Bernardo Gabriele Collaco, Cesar Abraham Gomez-Cabello, Syed Ali Haider, Ariana Genovese, Zhihui Fang, Nadia Wood, Sanjay Bagaria, Cui Tao, Antonio Jorge Forte
Cite
APA
Prabha, S., Collaco, B. G., Gomez-Cabello, C. A., Haider, S. A., Genovese, A., Fang, Z., Wood, N., Bagaria, S., Tao, C., & Forte, A. J. (2026). Fine-Tuning, Retrieval-Augmented Generation, and Hybrid Large Language Models for Postoperative Decision Support: Comparative Analysis. Journal of Medical Internet Research, 28, e90692-e90692. https://doi.org/10.2196/90692
Vancouver
Prabha S, Collaco BG, Gomez-Cabello CA, Haider SA, Genovese A, Fang Z, et al. Fine-Tuning, Retrieval-Augmented Generation, and Hybrid Large Language Models for Postoperative Decision Support: Comparative Analysis. Journal of Medical Internet Research. 2026;28:e90692-e90692. doi:10.2196/90692
BibTeX
@article{Prabha2026,
author = {Prabha, Srinivasagam and Collaco, Bernardo Gabriele and Gomez-Cabello, Cesar Abraham and Haider, Syed Ali and Genovese, Ariana and Fang, Zhihui and Wood, Nadia and Bagaria, Sanjay and Tao, Cui and Forte, Antonio Jorge},
title = {Fine-Tuning, Retrieval-Augmented Generation, and Hybrid Large Language Models for Postoperative Decision Support: Comparative Analysis},
journal = {Journal of Medical Internet Research},
year = {2026},
volume = {28},
pages = {e90692--e90692},
doi = {10.2196/90692}
}
EUVIMED