Related Experiment Video
Updated: Jul 19, 2026
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
An Intelligent IoT-Compatible Arrhythmia Detection System Using a Hybrid Vision Transformer-LSTM Framework
S Selva Birunda1, V Vaissnave2, K Abirami3
1Department of Artificial Intelligence and Data Science, Ramco Institute of Technology, Rajapalayam, India. selvabirunda89@gmail.com.
Cardiovascular Engineering and Technology
|July 8, 2026
Summary
A new hybrid deep learning model combining Long Short-Term Memory (LSTM) and Vision Transformer (ViT) accurately classifies electrocardiogram (ECG) signals for arrhythmia detection. This advanced system achieves 99.17% accuracy, proving effective for IoT healthcare applications.
Area of Science:
- Cardiology
- Artificial Intelligence
- Biomedical Engineering
Background:
- The Internet of Things (IoT) enables advanced medical signal processing for electrocardiogram (ECG) signal classification.
- Detecting and classifying arrhythmias from large, complex ECG datasets remains a significant challenge.
Purpose of the Study:
- To propose a hybrid deep learning (DL) approach combining Long Short-Term Memory (LSTM) and Vision Transformer (ViT) for automatic arrhythmia classification from ECG signals.
- To develop an efficient and accurate system for real-time ECG analysis in IoT-compatible healthcare settings.
Main Methods:
- Utilized the MIT-BIH database for training the DL model.
- Applied preprocessing techniques including min-max normalization and Stockwell transform for signal-to-spectrogram conversion.
- Employed the K-Means centered Adaptive Synthetic Sampling (KMADASYN) technique to address data imbalance.
- Implemented a Hybrid Optimized ViT with Long Short-Term Memory (HOVLSTM) for extracting spatial and temporal features.
Main Results:
- The proposed HOVLSTM model achieved a classification accuracy of 99.17%.
- The model demonstrated superior performance compared to existing ECG classification systems.
Conclusions:
- The developed technique is well-suited for IoT-compatible healthcare systems.
- The model's evaluation in a simulated IoT environment highlights its potential for future clinical decision-support systems.
