ARtificial intelligence to Enable Automated REspiratory illness Surveillance through Primary care (AREA-RESP)

COVID-19 has demonstrated the urgent need to detect and monitor emergent respiratory illnesses of epidemic or pandemic potential. In part 1, we will develop and evaluate algorithms that process primary care electronic medical record (EMR) data for case detection of acute respiratory illness. We will take advantage of a unique dataset to train models employing artificial intelligence (AI), where clinicians have diagnosed (or not) respiratory illness in 11,287 encounters, as part of the Public Health Agency of Canada (PHAC) FluWatch program. In part 2, we will go on to evaluate implementation of the most accurate approach for automated respiratory surveillance in practices that are part of a network of all seven practice-based research networks in Ontario (>1.5 million patients). Our study team includes experts in primary care, respirology, infectious diseases, EMR data, public health, epidemiology and computer science. Working with knowledge users, AREA-RESP will provide a foundation for national, representative population-based, real-time surveillance systems to monitor respiratory illness. Our work directly aligns with the Chief Public Health Officer priorities for public health system transformation to improve pandemic readiness, to leverage foresight tools, and to leverage big data and technology.