Grant: Collaborative Health Research Projects (NSERC partnered)

Pulmonary function (PF) tests are useful tools to diagnose and monitor lung disease. The gold standard test uses spirometry to measure FEV1 (forced expiratory volume in 1 sec). However, FEV1 only measures the large airways although many lung diseases begin in the small airways (SAw). If we could measure SAw function accurately, we could detect and treat lung diseases earlier. Airway oscillometry (OSc) measures SAw function accurately but OSc outputs are complex and difficult to interpret. For these reasons, OSc is not commonly used. To improve the clinical utility of OSc, we will develop machine learning (ML) to classify OSc outputs as normal or SAw obstruction. ML will 'learn' from spirometry and OSc data of bone marrow and lung transplant patients who undergo frequent PF testing because SAw obstruction develops often post-transplant. A large dataset with a wide range of normal/abnormal lung function values are important features that will allow ML to develop with high predictive accuracy. We will then validate the ML algorithm in patients with confirmed SAw obstruction. We will also assess the wider applicability of OSc and ML in the general population in 2 community PF laboratories with diverse lung diseases, and a community cohort in Fort McMurray. The project is led by a Lung Specialist (Chow) and Engineer with expertise in ML (Valaee). Our partners are appropriate: University Health Network is a world leading centre in lung diseases. Thorasys, a Canadian company, developed the only Health Canada approved OSc device. Earlier diagnosis of lung disease using ML with OSc will improve the health of Canadians and decrease the overall health burden. The project contributes to human resource development by providing rich learning experiences for graduate students and cardiopulmonary function technologists in Medicine and Engineering. The project adds to the Canadian economy by contributing to the medical device development and manufacturing in Canada.