COVID-19 mortality prediction among vaccinated non-palliative patients at emergency departments

The COVID-19 pandemic has placed extraordinary demands on the healthcare system and, at times, created the need to ration resources. Healthcare workers face the problem of efficient and fair allocation of limited resources to save as many lives as possible. In times of crisis, they require criteria based on empirical evidence to direct efforts toward patients who would benefit the most. An accurate mortality prediction model is thus a valuable tool for them. Such a model provides an opportunity for transparent discussion about the goals of care. Many such models were developed during the early days of the pandemic with limited data, and the recent ones were developed without considering the effects of vaccines. Since the start of the vaccination campaign, 47.6% of reported deaths were among unvaccinated patients. So, it is crucial to consider the vaccination status of patients when predicting mortality risk. My goal is to develop a mortality prediction model among vaccinated and non-palliative patients to guide the allocation of scarce resources during any future surge. I propose to develop a model using the Canadian COVID-19 Emergency Department Rapid Response Network database, with more than 198,000 registered patients from 51 Emergency Departments across Canada. I plan to include confirmed COVID-19 patients who presented to the emergency department after December 2021. I will use machine learning techniques to develop and implement a model to be used in emergency departments. Healthcare workers could input information, such as age, sex, symptoms, vitals, comorbidities, and vaccination status, to calculate the patients' mortality risk. Access to the model helps them allocate resources efficiently and fairly and have a transparent discussion about the goals of care.