Ambulatory assessment of human circadian phase and related sleep disorders from heart rate variability and other non-invasive physiological measurements

The circadian system plays a key role in our well-being and its effects can be seen in many aspects of our lives. Our master circadian clock, located deep in our brain, influences various physiological processes such as heart rate, body temperature, energy metabolism, hormones, and sleep, to name a few. Misalignments in our circadian system can lead to both short-term and long-term problems. It is possible to willfully shift our internal clock through the use of light therapy by forcing shifts in either direction depending on the timing of the light exposure. However, in order to do this correctly and effectively, it is necessary to know the timing of the circadian clock with respect to our solar clock. This is known as the circadian phase. In this thesis we have created mathematical models which are able to estimate circadian phase based on non-invasively collected signal such as heart rate, light exposure, and activity levels. The models have been trained and tested on various populations including young, middle-aged, and insomniacs. Furthermore, the models have been tested on data collected from a smartwatch equipped with an optical heart rate sensor, providing a potential use case for the rapidly-growing smartwatch market. In addition, a distinctive heart rate variability pattern was observed in sleep onset insomniacs which could be used as a tool to complement clinical diagnosis of this sleep disorder, as well as provide insights into the causes of insomnia