Scholarlink

Recurrent Neural Network (RNN) has lately proven to be extremely effective for analysing time-series data. This paper focuses on the automated detection of four distinct classes of ECG beats utilizing Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM) networks and Bidirectional LSTM (Bi-LSTM). The signals are sourced from MIT-BIH Arrhythmia Database. To enhance data quality, noise reduction is performed using Discrete Wavelet Transform and the ECG beats are segmented for labelling purposes. The models are trained from the scratch using these processed ECG beats before undergoing testing. The results indicate that model developed using Bi-LSTM units performed better when compared with models developed using modified GRU and modified LSTM. An accuracy of 98.77% using GRU model, 99.29% using LSTM model and 99.78% using Bi-LSTM model was observed.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Novel Deep Learning Approach for Automated Cardiac Arrhythmia Detection Using RNNs

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Mohebbanaaz,
  • K. Maheswari,
  • L. Lakshmi Prasanna Kumar

摘要

Recurrent Neural Network (RNN) has lately proven to be extremely effective for analysing time-series data. This paper focuses on the automated detection of four distinct classes of ECG beats utilizing Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM) networks and Bidirectional LSTM (Bi-LSTM). The signals are sourced from MIT-BIH Arrhythmia Database. To enhance data quality, noise reduction is performed using Discrete Wavelet Transform and the ECG beats are segmented for labelling purposes. The models are trained from the scratch using these processed ECG beats before undergoing testing. The results indicate that model developed using Bi-LSTM units performed better when compared with models developed using modified GRU and modified LSTM. An accuracy of 98.77% using GRU model, 99.29% using LSTM model and 99.78% using Bi-LSTM model was observed.