Facial photogrammetry to predict obstructive sleep apnea in a diverse paediatric population

On behalf of their research team, Dr. Indra Narang (Clinician Scientist) and Jun Au (Research Fellow) at The Hospital for Sick Children (SickKids), and members of the Edwin S.H. Leong Centre for Healthy Children, provide an update on their study developing a predictive model for the diagnosis of obstructive sleep apnea (OSA) in children using facial photographs taken by a smartphone.

OSA is a common sleep disorder where a person’s upper airway is repeatedly blocked during sleep, causing breathing pauses. This leads to lower oxygen levels in the blood and disturbed sleep. Currently, diagnosis of OSA requires an overnight sleep test called a polysomnogram (PSG). However, in Canada, less than 15% of children with OSA symptoms get this test due to a shortage of sleep testing facilities. Consequently, many children remain undiagnosed and untreated, which can result in long-term effects like poor academic performance, high blood pressure, diabetes, and mental health issues.

Thanks to the support from the Leong Centre Catalyst Grant, we initiated this pilot study to enroll children suspected of having OSA who are scheduled for a sleep test at SickKids and a community-based sleep clinic. We take simple 2-D photos of their faces using smartphones and then create 3-D models of their facial surfaces through a process called facial photogrammetry. From these 3-D images, we identify and extract specific facial features that could be associated with OSA. We will then develop a model, called 3-D PHOSA model, that can predict the likelihood of OSA based on these facial features.

We are currently actively recruiting participants. With additional support from the CIHR Project Grant, we will expand the study by recruiting more children from multiple centres to improve the model’s accuracy and to conduct an external validation study to confirm its applicability at other locations. We partner with sites across various provinces, deliberately including populations that reflect the racial and ethnic diversity of Canada. Additionally, we collaborate with community-based sleep clinics to include a broader and more representative patient population, ensuring that the research findings are applicable to real-world clinical populations. This approach also guarantees that people from different socioeconomic and geographic backgrounds have equal opportunities to participate in research.

At the study end, we will host a workshop for stakeholders including decision makers, knowledge users, technology providers, community paediatricians and caregivers to foster dialogue, review and discuss the study findings and plans for the next steps. If this research succeeds, it could lead to the development of a mobile app that enables healthcare professionals to identify children at risk for OSA more easily and quickly, facilitating earlier diagnosis and treatment, and helping children avoid long-term health problems associated with untreated OSA. The 3-D PHOSA model has the potential to mitigate the shortage of diagnostic sleep services and address inequities in access and delivery of sleep services for children with OSA across Canada and worldwide. This could fundamentally change current clinical practice paradigms, making the 3-D PHOSA model an innovative tool capable of advancing equity in sleep health for all children.