跨医学体系下医疗知识图谱的构建与药物预测研究——以动脉粥样硬化为例  

Construction of Medical Knowledge Graph and Drug Prediction Research under the Cross Medical System:Taking Atherosclerosis as an Example

作　　者：张君冬 杨松桦 严颖 李娜 刘江峰 黄奇[1] Zhang Jundong

机构地区：[1]南京大学信息管理学院,江苏南京210023 [2]郑州大学计算机与人工智能学院,河南郑州450001

出　　处：《情报理论与实践》2024年第2期178-188,共11页Information Studies:Theory & Application

摘　　要：[目的/意义]探讨跨医学体系下的医疗知识图谱构建框架,促进医疗健康数据的有机融合和深层次互通。[方法/过程]以5个国内公开的大型医学数据库为数据来源,通过知识重组、实体对齐、关系融合构建医疗知识图谱,在此基础上,以动脉粥样硬化这一疾病为例,进行药物预测实证研究。[结果/结论]最终形成包含153849个实体、6569280个三元、73种语义关系的医疗知识图谱,为改善因医学体系差异性而导致的中西医“松散”联系现状做出了积极探索,提升了现有医疗知识图谱的知识服务范围,同时也为解析药物预测结果带来了新的研究思路。[Purpose/significance]Explore the framework for constructing a medical knowledge graph across medical systems,and promote the organic integration and deep interoperability of medical health data.[Method/process]Taking five domestic public large-scale medical databases as data sources,the medical knowledge graph was constructed through knowledge reorganization,entity alignment,and relationship fusion.On this basis,taking atherosclerosis as an example,the empirical study of drug prediction was carried out.[Result/conclusion]Finally,a medical knowledge graph containing 153849 entities,6569280 triads,and 73 semantic relationships was formed.This study made an active exploration to improve the current situation of the“loose”connection between Chinese and Western medicine due to the differences in the medical system,improved the knowledge service scope of the existing medical knowledge graph,and also brought new research ideas for analyzing drug prediction results.

关 键 词：跨医学体系 医疗知识图谱 术语关联 链路预测 药物预测 中西医结合 

分 类 号：G353.1[文化科学—情报学] R-05[医药卫生]