Cardiovascular Risk Assessment from Wrist Photoplethysmography Signal of Smart Watch Using Attention-Based Techniques

The development of VLSI technology has led to the prosperous growth of wearable devices. With the global challenge posed by an increasing aging population to medical and social care systems, wearable devices have emerged to enable pervasive healthcare and physiological assessment. In this project, we aim to develop technology to assist cardiovascular risk assessment including blood pressure (BP) estimation and pulse wave velocity (PWV) estimation from wrist photoplethysmography (PPG) gathered by smart watch. The attention-based techniques which allow the model to focus on different part of input sequence to reveal long-range dependencies will be employed to upgrade the estimation accuracy. The model efficiency is also considered for lightweight requirements of edge-computing applications.