Drowsiness Detection in Humans based on ECG Analysis Using Temporal Convolutional Network

The monitoring of vital signs in humans holds a significant importance in various applications such as surveillance of elderly, smart driving, remote health monitoring, and sleep medicine. Among diverse techniques, ML-based architectures dedicated for ECG analysis have shown advantageous results. In this study, we present a classification architecture based on convolution networks called Temporal Convolutional Network (TCN) to categorize ECG signals based on the three vigilance states of a person: awake, sleepy, and asleep. The proposed model utilizes a very low number of parameters and yet achieves impressive outcomes compared to state-of-art studies. This makes the proposed classification model an excellent choice for energyefficient embedded circuits for various purposes.