Sleep apnea occurs due to the intermittent collapses of pharyngeal airway during sleep which causes pauses in breathing during sleep, and increases the risk of heart disease, asthma, and greater healthcare utilization. My team has developed a wide range of machine learning algorithms to assess pharyngeal airway characteristics and diagnose sleep apnea based on signals that can be recorded conveniently at home, including chest movement, respiratory sounds, heart sounds, speech, and snoring. The pharynx plays a key role in many vital functions, including breathing, swallowing, and speech. The gold standard methods to assess pharyngeal airway physiology are expensive imaging modalities like MRI. In this talk, I will discuss how we have developed reliable and reproducible protocols based on ultrasonography to estimate pharyngeal airway narrowing in patients with sleep apnea. We expanded this work and developed machine learning algorithms based on speech and snoring sounds to assess pharyngeal airway physiology, such as area, length, tissue mass, and fluid content. This work can lead to the development of accessible technologies for acoustic imaging of the pharyngeal airway. In this talk, I will also present how we have developed physiology, we have developed physiology-driven machine learning algorithms that can automatically 1) detect sleep apnea, sleep stages, and respiratory airflow during sleep, 2) separate respiratory and heart sounds from snoring sounds, and 3) assess cardiac function during sleep. These algorithms were validated to assess sleep apnea, exercise capacity, drowsiness, asthma worsening, and respiratory depression with opioids. I will also briefly discuss deep learning algorithms based on infra-red video data during sleep to estimate respiration, heart rate and sleep apnea severity without any patient contact.

Dr. Yadollahi holds a Canada Research Chair-Tier 2 in Cardio-Respiratory Engineering, is a Senior Scientist at the University Health Network’s KITE research institute (UHN-KITE), an Associate Professor at the University of Toronto's Institute of Biomedical Engineering, and an adjunct faculty at the University of Manitoba. Dr. Yadollahi is Director of FabrIc-Based REsearch (FIBRE) platform, with the vision to deliver revolutionary textile-based wearables for providing equitable access to healthcare, wherever users are.

Dr. Yadollahi’s research is focused on developing digital technologies to provide equitable access to healthcare for people with cardio-respiratory disorders. At UHN-KITE, Dr. Yadollahi leads the SleepdB laboratory, which includes state-of-the-art technologies to examine the intricate interplay between body fluid and cardio-respiratory disorders. SleepdB has gold standard clinical equipment to assess sleep and cardio-respiratory function. Moreover, through special infrastructure that enables full control of lighting and acoustics, SleepdB can realistically simulate home or in-hospital environments for technology development and validation. Yadollahi is a strong advocate of inclusion, diversity, equity, and accessibility (IDEA), and chairs UHN Research’s IDEA committee. Beyond UHN, she leads and co-leads several national initiatives to train the next generation of researchers and innovators to design digital technologies to promote health equity and IDEA. 

To date, Dr. Yadollahi has authored and co-authored more than 80 peer-reviewed manuscripts, presented over 150 times in scientific conferences, filed 3 patents, and been invited to give 78 talks on her research at prominent national and international academic institutions.