After California issued a state-wide Shelter in Place order to help to curb the spread of COVID-19, we wanted to start a broader conversation about the need for long-term interventions beyond the first few weeks or months of the pandemic. We adapted a mathematical model to represent SARS-CoV-2 transmission through a population of people who are susceptible; infected but pre-symptomatic, asymptomatic, mildly symptomatic, or hospitalized; or recovered. We used model parameter estimates collated and made publicly available through the MIDAS research consortium to make modeled dynamics roughly match what we have observed so far. We explored a few key scenarios, and created an online user interface for people to try out different possible long-term intervention strategies for themselves. The website is available here: covid-measures.github.io. The key results so far are:
Social distancing has two benefits: it flattens the curve to try to avoid overwhelming our healthcare systems (#FlattenTheCurve), and it delays the peak to buy time for our treatment capacity and resources to improve (#DelayThePeak).
Unfortunately, a short initial phase of social distancing will not be enough. If we lift the intervention completely, we expect to see a resurgence of disease.
There are several long-term strategies to avoid this resurgence (#KeepItFlat). One is the “light-switch” method, in which we use surveillance of hospitalized COVID-19 patients to turn social distancing measures on and off over the next 12-18 months, as we await better treatments or an effective and widely available vaccine.
We encourage you to play around with the model yourself, and see how the combination of first-wave and second-wave intervention timing, duration, and intensity affects the predicted epidemic dynamics. Please note that our aim here is to gain a qualitative understanding of possible scenarios, not to make quantitative predictions, which would require a far more detailed model. All of our code is openly available here.
I am deeply grateful to the talented and selfless team from my lab group and around Stanford who committed their days, nights, and weekends to this project: Marissa Childs, Morgan Kain, Devin Kirk, Mallory Harris, Jacob Ritchie, Lisa Couper, Isabel Delwel, Nicole Nova. I’m also grateful to our funding sources: NSF EEID (DEB-1518681), NIH MIRA (1R35GM133439-01), the Helman Scholarship, the Terman Award, the Stanford Data Science Scholarship, and the Terry Winograd Fellowship.
