Key result
Attention-enhanced hybrid CNN-LSTM model achieves ~99% accuracy for real-time ECG arrhythmia detection.
Why the study?
Early detection of cardiac arrhythmias using wearable devices is challenging because conventional cloud-based deep learning methods have significant latency and raise privacy concerns from transferring sensitive health data.
Does an attention-enhanced hybrid CNN-LSTM model in a federated fog computing environment accurately classify ECG arrhythmias while reducing latency and preserving privacy?
Does an attention-enhanced hybrid CNN-LSTM model in a federated fog computing environment accurately classify ECG arrhythmias while reducing latency and preserving privacy?
A novel federated fog computing framework using an attention-enhanced hybrid CNN-LSTM model provides highly accurate (99.06%) and low-latency real-time ECG arrhythmia classification while preserving data privacy.
Should not yet change wearable arrhythmia monitoring practice; leaves open real-world utility pending prospective clinical validation.
Early detection of cardiac rhythm disorders (arrhythmias) poses a significant challenge, especially when using wearable IoT devices that continuously monitor ECG signals. Conventional cloud-based deep learning methods are largely infeasible due to considerable latency and raise serious data privacy concerns by transferring sensitive health information to central servers. In this paper, we propose a novel system using federated fog computing for real-time arrhythmia detection and classification while protecting patient data privacy. This proposed system deploys compact deep learning (DL) models on proximate fog nodes, such as smartphones or home gateways, in close proximity to the user. Instead of sending raw ECG data, models are trained locally, and only model updates are sent to a cloud server for aggregation to avoid data leakage and reduce latency. We create a lightweight hybrid model that combines an attention mechanism with convolutional neural networks (CNNs) and long short-term memory (LSTMs), specifically designed for low-power fog devices. Our proposed model achieves a classification accuracy rate of 99.06, precision of 99.04, recall of 99.06, and a weighted average F1-score of 99.04 on the public MIT-BIH arrhythmia dataset. The experimental results show that the inference latency is significantly reduced to 0.22 ms (cleartext), 1.24 ms (encrypted) per sample/heartbeat inference time, and the encryption throughput reaches 577 beats/s using the homomorphic encryption technique. This approach is a suitable alternative for real-time ECG monitoring within a federated fog computing framework because it strikes a fair balance among speed, accuracy, and privacy.
No takes yet. Share an insight, caveat, or question.
Bhuyan et al. (2026) studied Arrhythmia (n=47). Attention Enhanced Hybrid CNN-LSTM Model in Federated Fog Computing vs. Centralized models was evaluated on Classification accuracy. The proposed attention-enhanced hybrid CNN-LSTM model in a federated fog computing framework achieved a classification accuracy of 99.06% for real-time ECG-based arrhythmia detection.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: