COVID-19 Variant Supplement – [email protected]: Development and validation of a multivariable prediction model of deterioration in patients diagnosed with COVID‑19 who are managing at home
Millions of Canadians are anticipated to be infected with COVID‑19 during this pandemic and many more will contract it in ongoing community transmission and/or a possible second wave. The majority of people who test positive for COVID‑19 are sent home to isolate. In this population, deterioration of their disease can happen quickly and without warning, and we currently cannot accurately predict the approximately 20% who deteriorate and need hospitalization. From discussions with our patients and patient advisor, we know that people who are isolating at home feel terrified and alone. We need an effective and safe ambulatory care and research strategy for people with COVID‑19 isolating at home. We are a team of heath care workers, patients, researchers and computer scientists (WearCOPD.ca; Can-BREATHE.ca) with five years of experience developing and using remote monitoring systems for respiratory disease. We have already built a smartphone application to facilitate the care of people with COVID‑19 at home by allowing them to report their symptoms to their physician. With this project, we will expand our system to also include continuous smartwatch-based monitoring of heart rate, respiratory rate, cough, speech and other parameters. Sensor data will provide us with large volumes of objective data and allow us to build accurate real time machine learning models for predicting who needs to go to hospital. We will integrate these models into a dashboard that alerts clinicians of any patients that area getting worse, so that they can be called into hospital. Patients can be reassured that they are being followed thoroughly even though they are at home. Our system will also provide a platform for further research into how to prevent long term sequalae and preserve the health of people with COVID‑19 who do not require hospitalization.