Spirometry is the gold standard medical test used to diagnose common and severe respiratory diseases such as asthma and chronic obstructive pulmonary disease (COPD). Diagnostic level spirometry requires that a patient execute a breathing maneuver in accordance with standard quality criteria; unfortunately, this maneuver is difficult for patients to perform, and specialty-level provider expertise is required to evaluate each maneuver for quality features and errors. The inability to determine the quality of spirometry at the point of care results in restricted access to high-quality testing and data, improper test interpretation, loss of confidence in and underutilization of spirometry, and underdiagnosis of disease. This UCSF and UC Berkeley faculty collaboration seeks to develop a novel machine learning solution to automate spirometry quality assurance. Automated spirometry quality assurance has great potential to dramatically expand access to specialty-level spirometry to all non-specialty care settings and for under resourced communities.
