医院人工智能算力平台建设研究与评估  

Research and Evaluation on the Construction of Hospital Artificial Intelligence Computing Power Platform

作　　者：王力华 郭铃琛 王鹏飞[2] 刘升平 唐立瑶 WANG Lihua;GUO Lingchen;WANG Pengfei;LIU Shengping;TANG Liyao(Beijing Friendship Hospital,Capital Medical University,Beijing 100050,China)

机构地区：[1]首都医科大学附属北京友谊医院,北京市100050 [2]华为技术有限公司,广东省深圳市518129 [3]云知声智能科技股份有限公司,北京市100096

出　　处：《中国卫生信息管理杂志》2025年第1期1-7,共7页Chinese Journal of Health Informatics and Management

基　　金：国家重点研发计划生物与信息融合(BT与IT融合)重点专项“基于知识增强的大规模认知智能模型关键技术研究”(2023YFF1204102)。

摘　　要：随着新一代信息技术的发展,卫生健康行业积极推进“人工智能+”应用创新。人工智能应用贯穿患者就医全流程,尤其是大模型的应用,需要强大算力支持,医院人工智能算力平台建设面临诸多挑战。本研究聚焦于医院算力平台建设,阐述算力基础设施组成与平台架构,结合理论与医院实践评估人工智能算力平台建设模式与建设规模。通过提出院内训推一体模式、云上集中训练边缘推理模式、全云化算力平台3种建设模式,测算出训练、推理的算力规模,为医院构建科学合理、高效可靠的人工智能算力平台提供理论与实践参考,推动医院数智化进程。With the development of the new generation of information technologies,the health industry has been actively promoting the application innovation of“artificial intelligence+”.Artificial intelligence runs through the entire process of patients’medical treatment.Especially with the application of large models,powerful computing power support is required.The construction of hospital artificial intelligence computing power platforms is facing numerous challenges.This study focuses on the construction of hospital computing power platforms,elaborates on the composition of computing power infrastructure and platform architectures,and evaluates the construction models and scales of artificial intelligence computing power platforms by combining theory with hospital practice.Three construction models are proposed,namely the in-hospital training and inference integration model,the cloud-based centralized training and edge inference model,and the fully cloud-based computing power platform.The computing power scales for training and inference are also calculated.It provides theoretical and practical references for hospitals to build scientific,rational,efficient and reliable artificial intelligence computing power platforms,and promotes the digital and intelligent process of hospitals.

关 键 词：医疗人工智能 大模型 医院人工智能算力平台 算力估算 

分 类 号：R-39[医药卫生] R319