Excited to tell you all about a new paper re COVID19 from a big team effort w/ @abhishekn, @akbarpour_, @Pietro_Tebaldi, @Simon_Mongey, Cody Cook, Aude Marzuoli, Matteo Saccarola and Hanbin Yang
reopenmappingproject.com/files/network-…
reopenmappingproject.com/files/network-…
Tl;dr: Heterogeneity matters when thinking about lockdown/re-opening policies. Diffs in concentrations of places where ppl encounter each other, diffs in industry, demographic (and co-morbidity) distributions, diffs in when the virus hit.
These forms of heterogeneity are (largely) measurable -- and we made a major effort to measure them. We combine a representation of daily activities + meetings across metro areas built on rich cell data by Replica w/ electronic medical records, O*Net, OES + more
We embed a non-homogeneous graph model w/ age-industry-comorbidity-work-from-home-ability types into an SIR-style model that accounts for [a]symptomatic infections, quarantine and healthcare.
Infection rates + initial compartment sizes are estimated by indirect inference; the rest come from data we've computed, or from leading figures in the literature. Contact rates come from our rich synthetic population representations.
To simulate policies, we change our contact matrices: no school --> no contacts in schools; 60+ stay at home? no contacts for 60+s except minimal neighborhood interactions. Our synthetic population gives us a lot of flexibility, and we've only scratched the surface.
To match policy discourse, we plot a "frontier" of the # of deaths vs # of days of employment lost as a result of diff't re-opening policies in Chicago, Sacramento and NYC. We plot them in abs terms, and in % relation to "cautious reopening".
A few insights jump out: (1) Big diffs across metros. A "work from home (WFH) if possible" (and work at work o.w.) policy would lead to 40% fewer deaths in Chicago; ~20% fewer in NYC; almost no diff in Sacramento. Why? Diffs b/w metros + b/w disease paths before mid-May.
(2) Two policies stand out: WFH if possible and "alternating schedules" (AS), which splits work/school cohorts into 2 that never meet. WFH has lower unemployment, but AS has fewer deaths. Keeping 60+ at home appears Pareto dominated, however.
A few notes: (1) We focus on policies with "caution" as a baseline -- basic neighborhood interactions are kept at a minimum. "Full" reopening would be much worse. (2) We focus on "extreme" pts of policies for exposition, but ideally we'd combine them!
Much more to come, but we're keen for thoughtful feedback + suggestions. We'll post code (freely reproducible using FRED instead of Replica) soon + results across all MSAs -- watch for it here: reopenmappingproject.com
Thank you to the many people who have poured hours into making this project possible -- esp Mark Cullen + Suzanne Tamang at the Center for Population Health Sciences at Stanford University and all the hard working folks at Replica
@codyfcook has a twitter too!
