Deep Learning-Based Blood Pressure Measurement on a Low-Power Microcontroller

Blood pressure is one of the most vital signals characterizing the health status of the human body. Based on this physiological data, many current and future diseases can be detected. Therefore, having a device that can continuously measure blood pressure without a cuff or medical intervention would be highly beneficial. This study aims to investigate non-invasive blood pressure measurement with the help of a low-power microcontroller, namely STM32F446RE. These measuring methods are based on deep neural networks and utilize externally measured photoplethysmography (PPG) and electrocardiogram (ECG) signals. The performance of the proposed models is compared to similar solutions from the literature and inspected for memory usage and runtime. The results of this work show that a convolutional layer-based model can be used for proper blood pressure estimation in such an energy-efficient device.