텐서플로 라이트를 이용한 엣지 컴퓨팅 기반의 휴대용 심전도 모니터

초록

In recent years, health concerns continue to grow, reflected in the growth of the healthcare and wellness industry. In particular, the importance of early detection and prevention of heart-related diseases is emphasized, development of ECG monitoring technology is required. This paper presents application for electrocardiogram (ECG) monitor as edge device implemented by Arduino Nano 33 BLE. In general, IoT devices utilizing microcontrollers have limited resources, avoid deploying software for machine learning inference on edge devices. Therefore, the system attempted to make high-performance machine learning inference on low-power embedded processors using the TensorFlow Lite for Microcontroller framework running on microcontrollers (MCU) to support real-time machine learning (ML) inference on low-power and resource-constrained edge devices. The performance of the pre-trained model and the lightweight model was compared through TensorFlow Lite to confirm whether the application was suitable.