II导联心电图中心肌梗死检测与定位：基于多尺度残差模块融合改进通道注意力模型

打开文本图片集

A myocardial infarction detection and localization model based on multi-scale field residual blocks fusion with modified channel attention

WU Qiucen，LUXueqi，WENYaoqi，HONGYong1，WUYuliang²，CHENaomin

choolfoli

HospitalofSouthernMedicalUniversity（DongguanPeople'sHospital），ongguan523o59，China

Abstract：Objective Weproposeamyocardialinfarction （MI）detectionandlocalizationmodelfor improving the diagnostic accuracyforMItoprovideasistance toclinicaldecision-making.Methods The proposed model was constructed basedon multi-scalefieldresidualblocksfusionmodifiedchannelattention（MSF-RB-MCA）.ThemodelutilizesleadII electrocardiogam （ECG）signalstodetectandlocaizeMI，andextractsdiferentlevelsoffeatureinformationthroughthe multi-scalefieldresidualblock.Amodifiedchannelatentionforautomaticadjustmentofthefeatureweightswasintroduced toenhancethemodel'sabilitytofocusontheMIregion，therebyimprovingtheaccuracyofMdetectionandlocalization. ResultsA5-foldcros-validationtestofthemodelwasperformedusingthepubliclyavailablePhysikalisch-Technische Bundesanstalt （PTB） dataset.For MI detection，themodelachievedanaccuracyof 99.96% onthe test setwith aspecificityof 99.84% and a sensitivity of 99.99% .For MI localization， the accuracy， specificity and sensitivity were 99.81% 99.98% and 99.65%， respectively.Theperformances of the model forMI detection and localizationweresuperior tothoseof other comparison models.Conclusion The proposed MSF-RB-MCA model shows excelent performance inAI detection and localizationbased onlead IIECGsignals，demonstrating its great potential for application inwearable devices. Kevworde： mvocardial infarctinn： deon learnino： multi-ccale： racidal hlnck： modified channal attantion

根据世界卫生组织（WHO）的数据显示，随着年龄的增长，因心血管疾病（CVD）引发的死亡人数显著上升，其中大多数发生在老年人群中[。（剩余19972字）