现代信息科技2025，Vol.9Issue(14)：21-26,31,7.DOI:10.19850/j.cnki.2096-4706.2025.14.005
LS-DDI:融合LSTM和Self-Attention的药物-药物相互作用预测研究
LS-DDI:Prediction Research on Drug-Drug Interaction Integrating LSTM and Self-Attention
摘要
Abstract
The combined use of multiple drugs may lead to adverse drug reactions,causing health issues.Therefore,predicting potential Drug-Drug Interactions is crucial.This paper proposes an algorithm called LS-DDI,which integrates Long Short-Term Memory(LSTM)networks and the Self-Attention Mechanism for DDI prediction.Gaussian similarity is used to calculate three distinct features of substructures,targets,and enzymes to form the similarity matrices.LSTM extracts contextual information,while the Self-Attention assigns different weights to the three features,leading to the final prediction research.Through the five-fold cross-validation,experimental results on two different datasets demonstrate that LS-DDI outperforms four other comparison models,proving its superior performance.Finally,case studies involving the three drugs of Torasemide,Cannabidiol,and Dexamethasone validate the effectiveness of the proposed model in predicting unknown DDIs.关键词
长短期记忆网络/自注意力机制/药物相互作用/药物不良反应Key words
Long Short-Term Memory(LSTM)/Self-Attention Mechanism/Drug-Drug Interaction(DDI)/adverse drug reaction分类
信息技术与安全科学引用本文复制引用
陈星鑫,聂斌,苗震,杨洋..LS-DDI:融合LSTM和Self-Attention的药物-药物相互作用预测研究[J].现代信息科技,2025,9(14):21-26,31,7.基金项目
国家自然科学基金(82260849,61562045)江西中医药大学星火传承培养计划(2252000111) （82260849,61562045）
江西省教育厅科学技术项目(GJJ2200925) （GJJ2200925）
江西省中医药管理局科技计划项目(2020B0412) （2020B0412）
江西中医药大学校级科技创新团队发展计划(CXTD22015) （CXTD22015）
