SHI Rui 1 , ZHENG Bing 1 , YAO Xun 1 , YANG Hao 1 , YANG Xuchen 1 , ZHANG Siyuan 1 , WANG Zhenwu 1 , LIU Dongfeng 1 , DONG Jing 1 , XIE Jiaxi 1 , MA Hu 2 , HE Zhiyang 3 , JIANG Cheng 4 , QIAO Feng 5 , LUO Fengming 6 , HUANG Jin 7
  • 1. Department of Information, West China Hospital of Sichuan University, Chengdu, 610041, P. R. China;
  • 2. ICT Product Portfolio Management and Solution Department, Huawei Technologies Co., Ltd, Chengdu, 610041, P. R. China;
  • 3. iFLYTEK Healthcare Research Institute, iFLYTEK Healthcare Technology Co., Ltd, Hefei, 230031, P. R. China;
  • 4. Sichuan Branch, China Telecom Co., Ltd, Chengdu, 610041, P. R. China;
  • 5. Research and Development Department 1, China Mobile (Chengdu) Information and Communication Technology Co., Ltd, Chengdu, 610213, P. R. China;
  • 6. Department of Respiratory and Critical Care Medicine/Clinical Research Center for Respiratory Disease/Laboratory of Pulmonary Immunology and Inflammation/Department of High Altitude Medicine, Center for High Altitude Medicine, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China;
  • 7. Medical Device Regulatory Research and Evaluation Center/Sichuan Provincial Comprehensive Clinical Center for Public Health, West China Hospital of Sichuan University, Chengdu, 610041, P. R. China;
HUANG Jin, Email: huangjin@hotmail.com
Medical large model; data annotation; multimodal learning; medical record generation; articficial intelligence
10.7507/1007-4848.202503081
Export PDF Favorites Scan Get Citation

Abstract Full text Figures/Tables References Cited by

Objective  To construct large medical model named by "Huaxi HongYi"and explore its application effectiveness in assisting medical record generation. Methods  By the way of a full-chain medical large model construction paradigm of "data annotation - model training - scenario incubation", through strategies such as multimodal data fusion, domain adaptation training, and localization of hardware adaptation, "Huaxi HongYi" with 72 billion parameters was constructed. Combined with technologies such as speech recognition, knowledge graphs, and reinforcement learning, an application system for assisting in the generation of medical records was developed. Results Taking the assisted generation of discharge records as an example, in the pilot department, after using the application system, the average completion times of writing a medical records shortened (21 min vs. 5 min) with efficiency increased by 3.2 time, the accuracy rate of the model output reached 92.4%. Conclusion  It is feasible for medical institutions to build independently controllable medical large models and incubate various applications based on these models, providing a reference pathway for artificial intelligence development in similar institutions.

Citation： SHI Rui, ZHENG Bing, YAO Xun, YANG Hao, YANG Xuchen, ZHANG Siyuan, WANG Zhenwu, LIU Dongfeng, DONG Jing, XIE Jiaxi, MA Hu, HE Zhiyang, JIANG Cheng, QIAO Feng, LUO Fengming, HUANG Jin. Construction and application of the "Huaxi Hongyi" large medical model. Chinese Journal of Clinical Thoracic and Cardiovascular Surgery, 2025, 32(5): 587-593. doi: 10.7507/1007-4848.202503081 Copy

Copyright ? the editorial department of Chinese Journal of Clinical Thoracic and Cardiovascular Surgery of West China Medical Publisher. All rights reserved

  • Previous Article

    Expressions of p27Kip1, RalA, and SPOCK1 in elderly colorectal cancer patients and relation between their expressions and prognosis

  • Next Article

    Differentiation of pancreatic neuroendocrine tumors of different pathological grades using CT imaging features