以咳嗽为第一主诉的儿童疾病医疗数据共享模型构建  

Construction of Children Cough as Chief Complaint Data Sharing Model

作　　者：朱志文 齐国强 杨恺熠 赵永根[1,2,3] 余卓 俞刚 ZHU Zhiwen;QI Guoqiang;YANG Kaiyi;ZHAO Yonggen;YU Zhuo;YU Gang(Department of Data and Information,The Children’s Hospital,Zhejiang University School of Medicine,Hangzhou 310052,China;Sino-Finland Joint AI Laboratory for Child Health of Zhejiang Province,Hangzhou 310052,China;National Clinical Research Center for Child Health,Hangzhou 310052,China;Huiying Medical Technology(Beijing)Co.Ltd.,Beijing 100192,China)

机构地区：[1]浙江大学医学院附属儿童医院数据信息部,杭州310052 [2]浙江-芬兰儿童健康人工智能联合实验室,杭州310052 [3]国家儿童健康与疾病临床医学研究中心,杭州310052 [4]慧影医疗科技(北京)股份有限公司,北京100192

出　　处：《医学信息学杂志》2023年第3期64-68,共5页Journal of Medical Informatics

基　　金：国家重点研发计划“以咳嗽为第一主诉症状的儿童疾病智能分析平台及应用”(项目编号:2019YFE0126200);国家自然科学基金面上项目“基于多模态融合的儿童慢性咳嗽人工智能辅助诊断技术研究”(项目编号:62076218);浙江省公益技术应用研究项目“数字化改革中儿童医疗数据智能分级方法研究及示范应用”(项目编号:LGF22H180004)。

摘　　要：分析以咳嗽为第一主诉的儿童疾病背景及其医疗数据应用情况和发展困境,结合联邦学习技术提出多中心横向联邦学习的数据安全共享框架,介绍3类联邦学习基础概念,从框架实施目的、系统架构组成、应用流程以及系统优点等方面详细阐述模型框架及构建路径。The paper analyzes the background of children diseases with cough as chief complaint and the applications of its medical data as well as the dilemma of development.Combined with federated learning technology,it proposes a data security sharing framework of multi-center horizontal federated learning,introduces the basic concepts of three types of federal learning,and elaborates the model framework and construction path from the implementation purpose,system architecture composition,application process,system advantages,etc.

关 键 词：儿童咳嗽 横向联邦学习 数据安全 人工智能 

分 类 号：R-058[医药卫生]