基于注意力网络与知识图谱的药物靶点预测

打开文本图片集

（宁夏大学前沿交叉学院，宁夏银川750021）

中图分类号：TP311 文献标识码：A 文章编号：2096-4706（2026）05-0040-07

Drug Target Prediction Based on Attention Network and Knowledge Graph

HAI Qiang （School ofAdvanced InterdisciplinaryStudies，Ningxia University，Yinchuan 75oo21，China）

Abstract： Prediction of Drug-Target Interactions （DTIs） provides strong support for the analysis of drug mechanism of action.This paper presents a drug target prediction method based on Attention Network and medical Knowledge Graph. First，itestablssadugetricstructuredtextdatasetadonstructsamedicalKowledgeGahecond，itploa graphatentionmodelasanecodertotrainembeddingrepresentationsofntitiesandelations.Tird，itusesaconvolutional embeding modelasadecoder toachieve the medical multi-relationship prediction task.Experimentsshow that this method effectively discovers relations among drugs，targets and diseases and achieves superior prediction performance.

Keywords： drug target prediction; Knowledge Graph; attention; knowledge embedding

0 引言

在全球医药市场不断增长的背景下，尽管药物研发取得了一定进展，仍有大量患者未能及时获得有效治疗[]，药物研发效率低下，尤其在应对快速发展的疾病时，成为全球亟待解决的挑战之一。（剩余8998字）