The Quantum-Enhanced Convolutional Neural Network (QECNN) achieved a classification accuracy of 97.2% on the MIT-BIH dataset and 98.1% on the PTB-XL dataset for detecting abnormal ECG patterns.
Does a Quantum-Enhanced Convolutional Neural Network improve the accuracy and security of automated arrhythmia detection from ECGs compared to classical deep learning models?
A novel quantum-enhanced neural network framework demonstrates high accuracy and enhanced data security for automated ECG arrhythmia detection in medical IoT systems.
The proliferation of the Medical Internet of Things (MIoT) has emphasized the importance of electrocardiograms (ECGs) for automated arrhythmia detection. Conventional heartbeat classification approaches, relying on classical neural networks, face limitations such as dataset dependency, severe class imbalance, and risks to patient data privacy. Quantum computing offers a transformative paradigm for processing and analyzing ECG signals using quantum neural networks (QNN) and leverages quantum superposition and entanglement to enhance computational parallelism, representational capacity, and security. This study introduces the Quantum-Enhanced Convolutional Neural Network (QECNN) for efficient classification of abnormal ECG patterns in MIoT applications. To mitigate privacy concerns, a quantum blockchain framework is proposed, integrating quantum hash functions and quantum identity authentication to ensure the integrity and confidentiality of ECG data storage and transmission. Additionally, trained quantum convolutional kernels are developed to extract morphological features from ECG signals, facilitating accurate identification of abnormal heartbeat patterns. The experimental results show that the proposed QECNN-based approach achieves classification accuracy of 97.2%, precision of 96.8%, recall (sensitivity) of 95.9%, specificity of 97.5%, and an F1-score of 96.3% on MIT-BIH dataset. On PTB-XL dataset, QECNN-based approach achieves classification accuracy of 98.1%, precision of 97.6%, recall (sensitivity) of 96.9%, specificity of 98.4%, and an F1-score of 97.2%. This method outperforms both classical and advanced deep learning (DL) models in terms of robustness, reliability, and diagnostic performance. These findings demonstrate that the framework can accurately detect abnormal ECG patterns while keeping the false-alarm rate low. Rigorous security analysis confirms the resilience of the quantum blockchain framework against quantum computing-based attacks, providing a robust mechanism for safeguarding ECG data.
Al-Shamasneh et al. (Tue,) conducted a other in Arrhythmia (n=18,916). Quantum-Enhanced Convolutional Neural Network (QECNN) vs. Classical and advanced deep learning models was evaluated on Classification accuracy for abnormal ECG patterns. The Quantum-Enhanced Convolutional Neural Network (QECNN) achieved a classification accuracy of 97.2% on the MIT-BIH dataset and 98.1% on the PTB-XL dataset for detecting abnormal ECG patterns.