TinyML and edge intelligence applications in cardiovascular disease: A survey.

Area of Science:

• Edge computing
• Embedded systems
• Biomedical engineering

Background:

• Tiny machine learning (TinyML) and edge intelligence are crucial for resource-constrained devices.
• Wearable devices offer a platform for pervasive, low-power health monitoring.
• Cardiac anomalies require continuous monitoring and real-time analytics.

Purpose of the Study:

• To explore TinyML's potential in cardiovascular monitoring using wearable devices.
• To review TinyML enablers, networking solutions, and optimization techniques.
• To analyze deep neural networks for real-time ECG analytics on edge devices.

Main Methods:

• Overview of TinyML hardware and software enablers.
• Examination of Low-power Wide area network (LPWAN) for TinyML deployment.
• Discussion of knowledge distillation, quantization, and pruning for model optimization.
• Analysis of deep neural networks (CNNs, Autoencoders, DBNs, Transformers) for ECG data.

Main Results:

• TinyML facilitates efficient, low-power cardiovascular monitoring.
• Optimization techniques enable complex ML models on edge devices.
• Various neural network architectures show promise for ECG analytics.

Conclusions:

• TinyML is a transformative technology for real-time cardiovascular monitoring.
• Efficient deep neural networks are key for wearable cardiac anomaly detection.
• The integration of TinyML and edge intelligence advances healthcare technology.