You are here
Days-since-infection structured model for Covid-19 dynamics
COVID-19 Research Area(s): Diagnostics, Genomics & Transmission Dynamics, Healthcare Delivery & Policy
We have developed a dynamic model of covid-19 cases and hospitalization, and fitted the model to observed data for BC. The model can be used to evaluate alternative policies to limit infection rates, for planning purposes, a.o.
The biggest problem at present for model implementation and uncertainty for predictions is data: basic information about hospitalization and number of cases by risk groups is not available.
We have developed a dynamic model for Covid-19 dynamics that explicitly account for numbers of infected people by days since infection. This model preserves the complex time delay structure for infectivity, illness, and disease reporting. We partition the susceptible population into infection risk categories (sheltering population, essential workers, and health care workers) so as to account for continuing increase in infections in high risk categories even if social distancing has led to a high proportion of the overall population being relatively safe from infection. By distinguishing between actual and reported infection rates, we explicitly simulate the possibly large pool of infectious people who are not visible in the case data. We are refitting the model every day to reported case data, which will allow us to update key parameter estimates over time as more information becomes available about impacts of social distancing and other mitigation measures such as drug treatments. This adaptive refitting approach will allow us to avoid potentially misleading predictions that can result from statistical modeling approaches that makes hidden assumptions about changes in the susceptible population over time.Collaboration opportunities:
Model Intercomparison Project focused on protocol for dynamic model data, scenario development and evaluation