心电信号的分割方法可以有效地反映运动员的心脏功能和身体机能状况.通过人工对心电信号的手动分割往往耗费大量的时间和精力.为了实现自动化的心电信号分割,本文提出了一种改进的两层双向长短期记忆网络(bi-directional long short-ter...心电信号的分割方法可以有效地反映运动员的心脏功能和身体机能状况.通过人工对心电信号的手动分割往往耗费大量的时间和精力.为了实现自动化的心电信号分割,本文提出了一种改进的两层双向长短期记忆网络(bi-directional long short-term memory,BiLSTM)的心电图分割算法,可以前向和后向分析时间序列,以检测和定位重要波段,如P波、QRS波群和T波.实验使用公开QT数据集进行验证,以模拟运动员在赛前的心电数据.在与LSTM,BiLSTM以及两层BiLSTM的对比实验中,本方法的所有评价指标均有所提升.其准确率达95.68%,召回率为91.62%,精确度为91.05%,特异性为96.64%,F1分数为91.41%.结果表明该方法对心电信号进行分割具有较好的效果.展开更多
We investigate the Turing instability and pattern formation mechanism of a plant-wrack model with both self-diffusion and cross-diffusion terms.We first study the effect of self-diffusion on the stability of equilibri...We investigate the Turing instability and pattern formation mechanism of a plant-wrack model with both self-diffusion and cross-diffusion terms.We first study the effect of self-diffusion on the stability of equilibrium.We then derive the conditions for the occurrence of the Turing patterns induced by cross-diffusion based on self-diffusion stability.Next,we analyze the pattern selection by using the amplitude equation and obtain the exact parameter ranges of different types of patterns,including stripe patterns,hexagonal patterns and mixed states.Finally,numerical simulations confirm the theoretical results.展开更多
文摘心电信号的分割方法可以有效地反映运动员的心脏功能和身体机能状况.通过人工对心电信号的手动分割往往耗费大量的时间和精力.为了实现自动化的心电信号分割,本文提出了一种改进的两层双向长短期记忆网络(bi-directional long short-term memory,BiLSTM)的心电图分割算法,可以前向和后向分析时间序列,以检测和定位重要波段,如P波、QRS波群和T波.实验使用公开QT数据集进行验证,以模拟运动员在赛前的心电数据.在与LSTM,BiLSTM以及两层BiLSTM的对比实验中,本方法的所有评价指标均有所提升.其准确率达95.68%,召回率为91.62%,精确度为91.05%,特异性为96.64%,F1分数为91.41%.结果表明该方法对心电信号进行分割具有较好的效果.
基金the National Natural Science Foundation of China(Grant Nos.10971009,11771033,and12201046)Fundamental Research Funds for the Central Universities(Grant No.BLX201925)China Postdoctoral Science Foundation(Grant No.2020M670175)。
文摘We investigate the Turing instability and pattern formation mechanism of a plant-wrack model with both self-diffusion and cross-diffusion terms.We first study the effect of self-diffusion on the stability of equilibrium.We then derive the conditions for the occurrence of the Turing patterns induced by cross-diffusion based on self-diffusion stability.Next,we analyze the pattern selection by using the amplitude equation and obtain the exact parameter ranges of different types of patterns,including stripe patterns,hexagonal patterns and mixed states.Finally,numerical simulations confirm the theoretical results.