CNN-BiLSTM:一种基于光电容积脉搏波信号的房颤检测模型

打开文本图片集
中图分类号:TP18;TP391;R541 文献标识码:A 文章编号:1006-8228(2025)08-08-04
Abstract:Toexploretheomplexitymbedddinpulsewavesignalsandachieveeficientidentifiationofatrialfbilation(AF), thisstudyproposesadetectionmodelbasedonphotoplethysmography(PPG)signalsThemethodintegratesconvolutionalneural networks(CNN),bidirectionallongshort-termmemory(BiLSTM),andphotoplethysmography(PPG)technology.Afusedand optimizedCNN-BiLSTMmodelisconstructed.Comparedwithothermodels,thismethodachieveshighereficiencyindata preprocessing and greater sensitivity in AF detection,with superior performance in recall (92.89%) ,precision (92.48%) ,and F1 score (92.64%) .Theeffectivenessof thismethodisvalidated through comparison with various traditional models,indicating promisingaplicationpotential.Thismethodiseasytouseandfficientindetection,andcanpartiallyreplacetraditionalECGbased detection methods,offering new insights for long-term monitoring and follow-up of AF patients.
Keywords:Atrial fibrillation;Photoplethysmography;CNN;BiLSTM
0引言
房颤(AF)是一种常见的心律失常,传统识别主要依赖心电图(ECG),但ECG存在操作复杂、便携性差、成本高等局限[-2]。(剩余4316字)