Related Experiment Video
Updated: Sep 15, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
4.1K
A deep learning approach for heart disease detection using a modified multiclass attention mechanism with BiLSTM
Umesh Kumar Lilhore1, Sarita Simaiya1, Monish Khan2
1Department of Computer Science and Engineering, Galgotias University, Greater Noida, UP, India.
Scientific Reports
|July 12, 2025
Summary
A new Modified Multiclass Attention Mechanism with Deep Bidirectional Long Short-Term Memory (M2AM with Deep BiLSTM) improves heart disease diagnosis. This advanced model enhances classification accuracy and reduces misclassifications, offering a significant step forward in cardiovascular health.
Area of Science:
- Biomedical Engineering
- Artificial Intelligence in Healthcare
- Cardiology
Background:
- Heart disease is a leading global cause of death, often due to delayed diagnosis and classification challenges.
- Traditional machine learning and deep learning methods face limitations including misclassification, feature overlap, insufficient data, computational intensity, and noise sensitivity.
Purpose of the Study:
- To introduce a novel Modified Multiclass Attention Mechanism based on Deep Bidirectional Long Short-Term Memory (M2AM with Deep BiLSTM) for improved heart disease classification.
- To address limitations of existing methods by enhancing feature representation and reducing noise, misclassification, and feature overlap.
Main Methods:
- Developed a novel M2AM with Deep BiLSTM model incorporating class-aware attention weights for dynamic feature focus.
- Utilized a large dataset (6000 samples, 14 features) from MIT-BIH and INCART databases.
- Applied an Improved Adaptive band-pass filter (IABPF) for noise reduction and signal enhancement, and wavelet transforms for accurate signal segmentation.
Main Results:
- The M2AM with Deep BiLSTM model achieved high performance: 98.82% accuracy, 97.20% precision, 98.34% recall, and 98.92% F-measure.
- Demonstrated superior performance compared to Classic Deep BiLSTM (SD-BiLSTM), Naive Bayes (NB), DNN-Taylos (DNNT), Multilayer perceptron (MLP-NN), and Convolutional Neural Network (CNN).
- Showcased significant noise reduction and enhanced signal quality through IABPF and wavelet transforms.
Conclusions:
- The proposed M2AM with Deep BiLSTM offers a robust solution for accurate heart disease diagnosis, overcoming key limitations of current methodologies.
- This advancement represents substantial progress in improving the accuracy and reliability of cardiovascular disease classification.
- The model's ability to handle complex patterns and noisy data signifies a major step towards more effective clinical decision support systems.
