Design and Implementation of Low Power Embedded System for OSA Monitoring

Abstract

Sleep disorders, and more specifically Obstructive Sleep Apnea (OSA), are becoming a major population health problem. Original overnight polysomnography is expensive, complicated and inapplicable to long term home monitoring. OSA is associated with hypertension, arrhythmia and cognitive decline. Therefore, there is a need for low-power wearable systems to reliably screen for OSA in the daily living environment as well as in a situation of eldercare. This thesis presents the design and implementation of a low power embedded system for OSA monitoring. The system incorporates 2 key components: an NRF52840 system on chip (SoC) for data processing and wireless data transmission and an LSM6DSOTR inertial measurement unit (IMU) for measuring the respiration amplitude. Together, these modules comprise a small wearable platform to collect physiological signals and transmit it via Bluetooth Low Energy (BLE) to other platforms for further analysis. The sleep is then converted into an Apnea-Hypopnea Index (AHI) which gets categorized into normal, mild, moderate or severe risk levels based on the number of events detected during sleep. The system is able to detect OSA and generate AHI values that agree with the values generated by commercial sleep monitoring devices. A power characterization test revealed that the system can go all night and process consistent respiratory signals.