A blockchain-enabled HeartCare framework using a lightweight binary neural network achieved an overall accuracy of 95.93% and a sensitivity of 95.90% for detecting four types of heart illnesses.
Does a blockchain-enabled HeartCare framework with a lightweight BNN accurately detect heart illnesses from single-lead ECGs in resource-constrained devices?
A blockchain-enabled framework with a lightweight binary neural network can accurately and securely detect heart illnesses from single-lead ECGs on resource-constrained devices.
Cardiovascular diseases (CVDs) are the primary cause of mortality worldwide. The healthcare sector in India currently shows promise for substantial changes, specifically in the utilization and importance of the Internet of Medical Things (IoMT). Edge computing is necessary to make the IoMT more scalable, portable, reliable, and responsive. Security and privacy concerns impede the development and deployment of IoMT devices. The technology of blockchain can resolve security and privacy concerns. In this work, we implement a lightweight binary neural network (BNN) in a Cortex-M4 microcontroller (MCU) to enable the detection of four different types of heart illnesses present in a single-lead electrocardiogram (ECG) signal, in addition to proposing a blockchain-enabled HeartCare framework. The end-user can identify ailments and subsequently disseminate ECG results to medical professionals via a privacy-preserving blockchain-enabled framework. To acquire the ECG signal, a reusable fabric electrode was proposed and successfully fabricated. Finally, the BNN model is being trained utilising ECG databases of patients from the Indian continent, in addition to other state-of-the-art databases. The post-deployment validation of the proposed framework was conducted rigorously in alignment with the ACC/AHA Guidelines, resulting in an overall accuracy of 95.93% and a sensitivity of 95.90% for our BNN model.
Borah et al. (Thu,) conducted a other in Cardiovascular diseases. Blockchain-enabled HeartCare framework with a lightweight binary neural network (BNN) was evaluated on Overall accuracy and sensitivity for detecting four different types of heart illnesses. A blockchain-enabled HeartCare framework using a lightweight binary neural network achieved an overall accuracy of 95.93% and a sensitivity of 95.90% for detecting four types of heart illnesses.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: