Using Machine Learning for Wearables to Understand the Association of Sleep with Future Morbidity - Doctor of Philosophy Thesis

Aug 2, 2026

Upload by Eliana

Sleep is essential to life and is structurally complex. Understanding how sleep relates to health and future morbidity relies largely on studies using self-report sleep diaries that capture subjective experience. Wearable-based machine learning offers a way to move beyond diary limitations by extracting patterns from objective signals. This thesis investigates how learning from wearable data can support predictive analysis of future health outcomes, aiming to improve the evidence connecting sleep behavior to morbidity risk.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages230

Reading Time9 h 40 min

AccessFree to download

Acknowledgements

Supervisors and DPhil program support

Research groups and key collaborators

Sports, internships, and personal support

Abstract

Sleep and current evidence base

Limitations of self-report diaries

Wearables and machine learning approach

Goal: predicting future morbidity

Using Machine Learning for Wearables to Understand the Association of Sleep with Future Morbidity Hang Yuan Keble College University of Oxford A thesis submitted for the degree of Doctor of Philosophy Trinity 2024 To my grandma, who nurtured me, and to my mom, who taught me how to see the light in darkness. Acknowledgements First, I would like to thank both of my amazing supervisors, Aiden Doherty and Simon Kyle, who gave me invaluable support on both academic and personal levels. Research sometimes is not just about doing the work but also about taking care of oneself. I am particularly grateful that they supported me in taking a break from work with my family after enduring 1.5 months of isolation for COVID quarantine when I ﬂew back home in 2021 . I would also like to thank all the amazing people that I have met during the course of my DPhil. I will give my gratitude to them in chronological order. In the ﬁrst year of the DPhil, I was fortunate to be part of the inaugural cohort of the EPSRC Centre for Doctoral Training (CDT) Program. Because of the exposure to a wide arrange of research topics in health, I could identify the kind of work that would interest me most intellectually in the following years. Obviously, I would like to thank all the program directors, Jim Davis, Tom Nicholas, Nina Hallowell, and Jens Rittscher. Another key person who made this program so special is Leema Chapman, who made sure that we were looked after. Nina and Leema gave us insights into the CDT program and were outspoken advocates for us students. It is unfortunate that Nina passed away recently but I am sure she would be proud of who I have become. I would also like to thank my other fellow CDT students, with whom I learnt and grew together, including Henrique Aguiar, Andres Tamm, Yifan Yu, Kangning Zhang and many more. During the second year of my DPhil, I joined the OxWearables and Experimental and Clinical Sleep Medicine groups. Through my interactions with these lovely people, I became a more independent thinker and a researcher with many skills. I would like to thank Shing Chan, Rosemary Walmsley, Andrew Creagh, Catherine Tong, Aiden Acquah, Alain Shreves, Scott Small, Laura Portas, Scott Small, Stefan Van Duijvenboden, Abram Schonfeldt, Lizzie Hill, and Rachel Sharman. I would like to highlight the conversations that I have had with Shing Chan and Rosemary Walmsley. Both were my go-to people if I encountered any technical diﬃculties or was simply feeling hopeless at times. Those light-hearted conversations really made the whole DPhil experience far more enriching. As my DPhil project developed, I also had much help from the collaborators elsewhere, some of whom contributed data they collected, and some others provided critical feedback for my work in progress. I would give my warm regards to my collaborators in Australia, Leicester and Pennsylvania: Peter Eastwood, Amy Reynolds, Kathleen Maddison, Tatiana Plekhanova, Philip Gehrman and others whom I might not be able to have direct contact with. Even though Amy and I only had a few encounters during our collaboration, she gave me strong support in how to put writing into perspective and helped me understand how sleep clinicians think. On a personal level, I met so many friends by playing tennis at the University. It is the sport that kept me sane during COVID. Thank you to my coaches, Matej Lukic and Jarlath Brine, who taught me to be a better player and master the mental game of tennis. Thank you to my fellow tennis mates, Renyuan Xu and Tiger Yuan and friends from OULTC, who I still managed to see on a regular basis during the lockdown in the UK for sports with social distancing rules. In the summer of 2022, I was fortunate to have taken a short departure from my study program to do an internship at Meta, London. I have to say it was avery intense 12-week program, but I learnt a huge deal about how to build machine learning systems at scale and the best pr

FAQs

Why does the thesis focus on sleep and future morbidity?

Sleep is essential but complex, and the research aims to strengthen the connection between sleep and later health outcomes, specifically future morbidity risk.

What limitation of traditional sleep research does the thesis address?

Much prior evidence relies on self-report sleep diaries, which capture subjective experience and are limited compared with objective measurements.

How do wearables and machine learning fit into the thesis approach?

The thesis explores using wearable data and machine learning to learn patterns from objective signals, supporting predictive analysis of future health outcomes.

Aug 2, 2026

Upload by Eliana