Copyright: ©Author(s) 2027. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) license. No commercial re-use. See permissions. Published by Baishideng Publishing Group Inc.
World J Gastroenterol. Jan 7, 2027; 33(1): 122422
Published online Jan 7, 2027. doi: 10.3748/wjg.122422
Retrieval-augmented generation models for predicting pathological upgrading in endoscopically resected gastric low-grade intraepithelial neoplasia: A dual-center evaluation
Qian Zhuang, Jin-Nian Cheng, Yu-Lu Zheng, Shan Wu, Yi-Min Chu, Lu Zhou, Jie Xia, Min Ning, Hai-Xia Peng, Xin-Jian Wan, Zhi-Xia Dong
Qian Zhuang, Yu-Lu Zheng, Shan Wu, Jie Xia, Min Ning, Xin-Jian Wan, Zhi-Xia Dong, Digestive Endoscopic Center, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai 200233, China
Jin-Nian Cheng, Yi-Min Chu, Lu Zhou, Hai-Xia Peng, Digestive Endoscopic Center, Tongren Hospital, Shanghai Jiaotong University School of Medicine, Shanghai 200336, China
Jin-Nian Cheng, Yi-Min Chu, Lu Zhou, Hai-Xia Peng, Shanghai Key Laboratory of Flexible Medical Robotics, Tongren Hospital, Institute of Medical Robotics, Shanghai Jiao Tong University, Shanghai 200336, China
Co-first authors: Qian Zhuang and Jin-Nian Cheng.
Co-corresponding authors: Xin-Jian Wan and Zhi-Xia Dong.
Author contributions: Zhuang Q and Cheng JN conceptualized and designed the study, drafted the manuscript and contributed equally as co-first authors; Zheng YL and Wu S reviewed the data analyses and contributed to the discussion; Xia J, Ning M, Zhou L, and Chu YM assisted in data collection; Zhuang Q performed the literature search and construction of the retrieval-augmented generation knowledge base; Wu S provided statistical expertise and editorial assistance; Peng HX, Wan XJ, and Dong ZX supervised the study; Wan XJ and Dong ZX contributed equally as co-corresponding authors; and all authors contributed substantially to the work and approved the final version for publication.
AI contribution statement: Portions of this manuscript were edited using AI tools (Kimi,
https://www.kimi.com/) solely for language refinement. The authors carefully reviewed and verified all AI-assisted outputs and take full responsibility for the scientific content of the manuscript.
Supported by the Shanghai Municipal Health Commission Health Industry Clinical Research Project of China, No. 20234Y0016; the National Key Research and Development Program of China, No. 2023YFC2413803 and No. 2023YFC2413806; the Shanghai Municipal Health Commission Smart Healthcare Special Research Project of China, No. 2025ZHYL021; the Emerging Frontier Project of the Shenkang Development Center, China, No. SHDC12024124; and the Key Discipline Project of the Shanghai Municipal Health System, China, No. 2024ZDXK0004.
Institutional review board statement: This study was conducted following the Helsinki Declaration and was approved by the Ethics Committee of Shanghai Sixth People's Hospital [Approval No. 2026-KY-092(K)] and Ethics Committee of Shanghai Tongren Hospital (Tongren Lunshen K2026-031-01).
Informed consent statement: Patients were not required to provide informed consent to the study because the analysis used anonymous clinical data that were obtained after each patient agreed to endoscopic submucosal dissection by written consent.
Conflict-of-interest statement: The authors declare no competing financial interests.
Data sharing statement: The datasets generated and analyzed during this study are not publicly available due to patient privacy regulations but are available from the corresponding author upon reasonable request.
Corresponding author: Zhi-Xia Dong, MD, Chief Physician, Professor, Digestive Endoscopic Center, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, No. 600 Yi Shan Road, Shanghai 200233, China.
dzhixia2013@163.com
Received: April 20, 2026
Revised: June 15, 2026
Accepted: June 30, 2026
Published online: January 7, 2027
Processing time: 205 Days and 14.3 Hours
BACKGROUND
Gastric low-grade intraepithelial neoplasia (LGIN) exhibits highly variable clinical outcomes. The discordance between histological findings from endoscopic forceps biopsy and endoscopic submucosal dissection (ESD) complicates the formulation of appropriate treatment strategies.
AIM
To evaluate a prediction framework using large language models (LLMs) enhanced with retrieval-augmented generation (RAG) for the pathological upgrading of gastric LGIN.
METHODS
We retrospectively enrolled 329 patients with 372 gastric LGIN lesions from two tertiary hospitals who underwent ESD following a pre-operative diagnosis of gastric LGIN via biopsy. A RAG knowledge base was constructed based on 28 peer-reviewed articles on LGIN upgrading. Two LLMs (DeepSeek-V3 and Qwen3-235b) were evaluated in a zero-shot setting to generate upgrading probability predictions and reasoning reports for each lesion. Performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis (DCA). The performance of RAG-enhanced LLMs was compared with that of a generalized linear mixed model baseline and two experienced endoscopists. Subgroup analyses were also conducted.
RESULTS
Pathological upgrading to high-grade intraepithelial neoplasia or early gastric cancer was confirmed in 148 (40%) lesions. The AUCs for the RAG-enhanced DeepSeek-V3 and Qwen3-235b were 0.873 [95% confidence interval (95%CI): 0.828-0.901] and 0.862 (95%CI: 0.818-0.893), respectively, demonstrating performance comparable to that of experienced endoscopists and baseline model (DeLong test, P > 0.05). Based on the maximum Youden index, DeepSeek-V3 achieved a sensitivity of 85.1%, specificity of 80.4%, positive predictive value of 74.1%, and negative predictive value of 89.1% at the optimal threshold of 0.357; Qwen3-235b achieved 84.5%, 77.2%, 71.0%, and 88.3%, respectively, at the threshold of 0.453. DCA demonstrated superior clinical utility across threshold probabilities of 15%-80%. In the subgroup analysis, RAG-enhanced LLMs maintained higher accuracy in lesions measuring 1-2 cm.
CONCLUSION
This study evaluated a novel RAG-enhanced framework using LLMs for predicting pathological upgrading in gastric LGIN. These models may have the potential to support clinical decision-making for ESD, pending prospective validation.
Core Tip: This study evaluated a novel retrieval-augmented generation (RAG) -enhanced large language model framework using dual-center data to predict pathological upgrading in gastric low-grade intraepithelial neoplasia. The RAG-enhanced DeepSeek-V3 and Qwen3-235b models achieved areas under the curve of 0.873 [95% confidence interval (95%CI): 0.828-0.901] and 0.862 (95%CI: 0.818-0.893), respectively, demonstrating performance comparable to experienced endoscopists and confirming their potential clinical utility.
