PDF(683 KB)
PDF(683 KB)
PDF(683 KB)
医疗大模型技术及应用发展研究
Technology and Application Development of Medical Foundation Model
医疗大模型基于深度神经网络架构进行复杂医疗数据的高效处理与模式识别，为智慧医疗提供新型的决策支持；需要系统分析医疗大模型技术及应用情况，以精准把握医疗大模型的发展方向、精准应对面临的发展挑战，进而基于医疗大模型提升医疗文本、医学图像、药械研发、医学教育等方面的能力。本文梳理了医疗大模型的技术范式与应用场景，剖析了由基础层、模型层、应用层、公共模块构成的医疗大模型技术体系，覆盖评价指标体系构建、数据集范围与题型、模型对齐方法、模型评测平台的医疗大模型评测体系，辨识出医疗大模型应用存在的数据安全、技术风险、落地挑战、伦理道德等方面的难点。为此建议，发挥政府引导优势、保障数据安全，加快基础理论研究、突破技术风险，强化应用场景牵引、缓解落地挑战，建立健全监管机制、规范伦理道德，完善公共服务体系、营造创新生态，以加快医疗大模型创新应用，推动我国智慧医疗的高端化、智能化、绿色化发展。
The medical foundation model performs efficient processing and pattern recognition of complex medical data based on a deep neural network architecture, providing a new type of decision support for intelligent medical care. It is necessary to systematically analyze the technologies and application of the medical foundation model, thus to identify the development directions and challenges and improve the capabilities of the medical care sector in medical text writing, medical image recognition, medical equipment research and development, and medical education using the medical foundation model. This study sorts out the technology paradigm and application scenarios of the medical foundation model and proposes a technology system and an evaluation system for the model. The technology system is composed of a base layer, a model layer, an application layer, and a common module. The evaluation system involves the establishment of an evaluation index system, dataset range and question types, model alignment methods, and model evaluation platforms. Moreover, application challenges of the medical foundation model are identified in terms of data security, technical risks, implementing challenges, and ethics. Furthermore, the following countermeasures are suggested: (1) ensuring data security through government guidance, (2) accelerating basic theoretical research to address technic risks, (3) focusing on application scenarios to cope with implementing challenges, (4) improving the ethics regulating mechanism, and (5) perfecting the public service system to create an innovation ecosystem, thereby accelerating the innovative development of the medical foundation model and promoting the high-end, intelligent, and green development of intelligent medical care in China.
医疗大模型 / 多模态数据 / 预训练微调 / 提示工程 / 技术体系 / 评测体系
medical foundation model / multimodal data / pre-training and fine-tuning / prompt engineering / technology system / evaluation system
| [1] |
陈晓红, 许冠英, 徐雪松, 等. 我国算力服务体系构建及路径研究 [J]. 中国工程科学, 2023, 25(6): 49‒60.
Chen X H, Xu G Y, Xu X S, et al. Computing power service system of China and its development path [J]. Strategic Study of CAE, 2023, 25(6): 49‒60.
|
| [2] |
人工智能医疗器械创新合作平台, 中国信息通信研究院. 人工智能大模型赋能医疗健康产业白皮书 (2023) [R]. 北京: 人工智能医疗器械创新合作平台, 中国信息通信研究院, 2023.
Artificial Intelligent Medical Device Innovation and Cooperation Platform, China Academy of Information and Communications Technology. White paper on empowering the healthcare industry with AI large language models (2023) [R]. Beijing: Artificial Intelligent Medical Device Innovation and Cooperation Platform, China Academy of Information and Communications Technology, 2023.
|
| [3] |
亿欧智库. 2023医疗健康AI大模型行业研究报告 [R]. 北京: 亿欧智库, 2023.
EO Intelligence. Research report healthcare AI large language model industry (2023) [R]. Beijing: EO Intelligence, 2023.
|
| [4] |
陈晓红, 陈姣龙, 胡东滨, 等. 面向环境司法智能审判场景的人工智能大模型应用探讨 [J]. 中国工程科学, 2024, 26(1): 190‒201.
Chen X H, Chen J L, Hu D B, et al. Application of artificial intelligence large language model for smart environmental judicial adjudication [J]. Strategic Study of CAE, 2024, 26(1): 190‒201.
|
| [5] |
Wang H F, Li J W, Wu H, et al. Pre-trained language models and their applications [J]. Engineering, 2023, 25: 51‒65.
|
| [6] |
车万翔, 窦志成, 冯岩松, 等. 大模型时代的自然语言处理: 挑战、机遇与发展 [J]. 中国科学: 信息科学, 2023, 53(9): 1645‒1687.
Che W X, Dou Z C, Feng Y S, et al. Towards a comprehensive understanding of the impact of large language models on natural language processing: Challenges, opportunities and future directions [J]. Scientia Sinica Informationis, 2023, 53(9): 1645‒1687.
|
| [7] |
Singhal K, Azizi S, Tu T, et al. Large language models encode clinical knowledge [J]. Nature, 2023, 620(7972): 172‒180.
|
| [8] |
Thirunavukarasu A J, Ting D S J, Elangovan K, et al. Large language models in medicine [J]. Nature Medicine, 2023, 29(8): 1930‒1940.
|
| [9] |
Webster P. Six ways large language models are changing healthcare [J]. Nature Medicine, 2023, 29: 2969‒2971.
|
| [10] |
Moor M, Banerjee O, Abad Z S H, et al. Foundation models for generalist medical artificial intelligence [J]. Nature, 2023, 616(7956): 259‒265.
|
| [11] |
陈润生. 医疗大数据结合大语言模型的应用展望 [J]. 四川大学学报(医学版), 2023, 54(5): 855‒856.
Chen R S. Prospects for the application of healthcare big data combined with large language models [J]. Journal of Sichuan University (Medical Sciences), 2023, 54(5): 855‒856.
|
| [12] |
韩晓光, 朱小龙, 姜宇桢, 等. 人工智能与机器人辅助医学发展研究 [J]. 中国工程科学, 2023, 25(5): 43‒54.
Han X G, Zhu X L, Jiang Y Z, et al. Development strategies for artificial intelligence and robotics in medicine [J]. Strategic Study of CAE, 2023, 25(5): 43‒54.
|
| [13] |
颜见智, 何雨鑫, 骆子烨, 等. 生成式大语言模型在医疗领域的潜在典型应用与面临的挑战 [J]. 医学信息学杂志, 2023, 44(9): 23‒31.
Yan J Z, He Y X, Luo Z Y, et al. Generative large language lodels in the medical domain: Potential and typical applications and challenges [J]. Journal of Medical Intelligence, 2023, 44(9): 23‒31.
|
| [14] |
Huang Z, Bianchi F, Yuksekgonul M, et al. A visual-language foundation model for pathology image analysis using medical twitter [J]. Nature Medicine, 2023, 29(9): 2307.
|
| [15] |
Veen D V, Uden C V, Blankemeier L, et al. Adapted large language models can outperform medical experts in clinical text summarization [J]. Nature Medicine, 2024, 30: 1134‒1142.
|
| [16] |
Zhang X M, Wu C Y, Zhang Y, et al. Knowledge-enhanced visual-language pre-training on chest radiology images [J]. Nature Communications, 2023, 14: 4542.
|
| [17] |
Yu G, Sun K, Xu C, et al. Accurate recognition of colorectal cancer with semi-supervised deep learning on pathological images [J]. Nature Communications, 2021, 12: 6311.
|
| [18] |
Zhou H J, Liu F L, Gu B Y, et al. A survey of large language models in medicine: Principles, applications, and challenges [EB/OL]. (2024-02-02)[2024-07-15]. https://arxiv.org/html/2311.05112v3.
|
| [19] |
杨善林, 丁帅, 顾东晓, 等. 医疗健康大数据驱动的知识发现与知识服务方法 [J]. 管理世界, 2022, 38(1): 219‒228.
Yang S L, Ding S, Gu D X, et al. Healthcare big data driven knowledge discovery and knowledge service approach [J]. Journal of Management World, 2022, 38(1): 219‒228.
|
| [20] |
蛋壳研究院. 2023数字智慧病理行业研究报告 [R]. 重庆: 蛋壳研究院, 2023.
VCBeat Research. Research report on digital intelligence pathology industry (2023) [R]. Chongqing: VCBeat Research, 2023.
|
| [21] |
郑永年. 如何科学地理解"新质生产力"? [J]. 中国科学院院刊, 2024, 39(5): 797‒803.
Zheng Y N. How to scientifically understand "new quality productivity" [J]. Bulletin of Chinese Academy of Sciences, 2024, 39(5): 797‒803.
|
| [22] |
周文, 许凌云. 论新质生产力: 内涵特征与重要着力点 [J]. 改革, 2023 (10): 1‒13.
Zhou W, Xu L Y. On new quality productivity: Connotative characteristics and important focus [J]. Reform, 2023 (10): 1‒13.
|
| [23] |
Bi K F, Xie L X, Zhang H H, et al. Accurate medium-range global weather forecasting with 3D neural networks [J]. Nature, 2023, 619(7970): 533‒538.
|
| [24] |
Romera-Paredes B, Barekatain M, Novikov A, et al. Mathematical discoveries from program search with large language models [J]. Nature, 2024, 625(7995): 468‒475.
|
| [25] |
Fei N Y, Lu Z W, Gao Y Z, et al. Towards artificial general intelligence via a multimodal foundation model [J]. Nature Communications, 2022, 13: 3094.
|
| [26] |
郭华源, 刘盼, 卢若谷, 等. 人工智能大模型医学应用研究 [J]. 中国科学: 生命科学, 2024, 54(3): 482‒506.
Guo H Y, Liu P, Lu R G, et al. Research on a massively large artificial intelligence model and its application in medicine [J]. Scientia Sinica Vitae, 2024, 54(3): 482‒506.
|
| [27] |
余艳, 张文, 熊飞宇, 等. 融合知识图谱与神经网络赋能数智化管理决策 [J]. 管理科学学报, 2023, 26(5): 231‒247.
Yu Y, Zhang W, Xiong F Y, et al. Fusion of knowledge graph and neural network to empower data-intelligence for management decisions [J]. Journal of Management Sciences in China, 2023, 26(5): 231‒247.
|
| [28] |
Dias A L, Rodrigues T. Large language models direct automated chemistry laboratory [J]. Nature, 2023, 624(7992): 530‒531.
|
| [29] |
Shanahan M, McDonell K, Reynolds L. Role play with large language models [J]. Nature, 2023, 623(7987): 493‒498.
|
| [30] |
Boiko D A, MacKnight R, Kline B, et al. Autonomous chemical research with large language models [J]. Nature, 2023, 624(7992): 570‒578.
|
| [31] |
Piktus A. Online tools help large language models to solve problems through reasoning [J]. Nature, 2023, 618(7965): 465‒466.
|
| [32] |
Zhen C Q, Shang Y L, Liu X Y, et al. A survey on knowledge-enhanced pre-trained language models [EB/OL]. (2022-12-27)[2024-07-15]. https://arxiv.org/abs/2212.13428.
|
| [33] |
Pan S R, Luo L H, Wang Y F, et al. Unifying large language models and knowledge graphs: A roadmap [J]. IEEE Transactions on Knowledge and Data Engineering, 2024, 36(7): 3580‒3599.
|
| [34] |
Yin S K, Fu C Y, Zhao S R, et al. A survey on multimodal large language models [EB/OL]. (2023-06-23)[2024-07-15]. https://arxiv.org/abs/2306.13549.
|
| [35] |
Ding N, Qin Y J, Yang G, et al. Parameter-efficient fine-tuning of large-scale pre-trained language models [J]. Nature Machine Intelligence, 2023, 5: 220‒235.
|
| [36] |
Arasteh S T, Han T Y, Lotfinia M, et al. Large language models streamline automated machine learning for clinical studies [J]. Nature Communications, 2024, 15: 1603.
|
/
| 〈 |
|
〉 |
