基于动态图神经网络的医学技术机会发现方法研究  

Study on Technology Opportunity Discovery Method in Medical Field Based on Dynamic Graph Neural Network

作　　者：陈逸菲 王龙超 王德超 李永洁 唐小利[1] CHEN Yifei;WANG Longchao;WANG Dechao;LI Yongjie;TANG Xiaoli(Institute of Medical Information/Library,Chinese Academy of Medical Sciences&Peking Union Medical College,Beijing 100005,China)

机构地区：[1]中国医学科学院/北京协和医学院医学信息研究所/图书馆,北京100005

出　　处：《医学信息学杂志》2025年第3期29-35,共7页Journal of Medical Informatics

基　　金：中国医学科学院医学与健康科技创新工程项目(项目编号:2021-I2M-1-033)。

摘　　要：目的/意义提出一种基于动态图神经网络的技术机会发现方法,为动脉粥样硬化相关技术创新提供智能决策支持。方法/过程从incoPat获取2004—2024年动脉粥样硬化领域专利数据,提取技术要素,构建包含技术要素的动态技术语义网络,提出动态图神经网络模型DynGNN-TOD,采用滑动时间窗对动脉粥样硬化相关专利数据进行划分,并通过对比实验验证模型性能。结果/结论 DynGNN-TOD模型的AUC-ROC达0.932,Accuracy@10为62.3%,相较于其他3种基线模型均有不同程度的提升,能够为技术机会发现提供可靠的预测支持。Purpose/Significance To propose a method of technical opportunity discovery based on dynamic graph neural network,so as to provide intelligent decision support for the innovation of atheromatosis-related technologies.Method/Process The patent data in the field of atherosclerosis from 2004 to 2024 are obtained from incoPat,the technical elements are extracted,and a dynamic technical semantic network containing the technical elements is constructed.A dynamic graph neural network model named DynGNN-TOD is proposed,and the atherosclerotic patent data are divided by sliding time window.The performance of the model is verified by comparative experiments.Result/Conclusion Experimental results show that the AUC ROC of DynGNN-TOD is 0.932 and Accuracy@10 is 62.3%,which are improved in different degrees compared with the other 3 baseline models,and can provide reliable prediction support for technical opportunity discovery.

关 键 词：技术机会发现 动态图神经网络 技术语义网络 技术组合进化理论 

分 类 号：R-058[医药卫生]