Digestive endoscopy in the era of COVID19: An opportunity for optimizing timely equitable care during the pandemic and beyond

Digestive endoscopies are amongst the most frequently practiced procedures, allowing for specialized endoscopic therapy and permit diagnosis and treatment of pre-cancerous lesions in the upper and lower GI tracts as well as further oncological management. The SARS-coV-2 pandemic has dramatically impacted on endoscopic volumes across Canada; most units saw a sudden reduction of almost 90% of endoscopies with a resulting unmanageable increase of postponed exams since the first pandemic wave hit that persisted even after activity resumption. Amalgamating data from five centers, a "best-case" scenario suggests 30,757 procedures (35% of usual yearly totals) will NOT be performed; with a worst-case scenario reaching 69,301 over the 12 months following the initial COVID outbreaks, depending on pandemic epidemiology, mitigation protocols, resources shortages, and patient apprehension about attending a procedure. The crisis is compounded by a paucity of evidence-based forecasting and few procedural priority setting instruments. We propose to establish a toolkit for GI endoscopy units across Canada, addressing these issues by developing and validating clinical and resource management instruments. We will create a dataset of almost 90,000 procedures, designing validated referral prediction for gastroscopies and colonoscopies, allowing for a credible, standardized and equitable approach for triaging these procedures into hierarchal groups. Data collection will be facilitated by implementing a national learning process-based AI platform, while predictive modelling will be validated employing machine learning. We will also use AI methods to optimize procedural scheduling that varies widely across the country. The proposed deliverables develop, test, and implement effective approaches to manage the consequences of COVID-19 at individual and population levels by creating generalizable instruments that amplify existing research platforms and AI infrastructure.