-
Notifications
You must be signed in to change notification settings - Fork 46
/
metadata-CU-AGE-ST.txt
8 lines (8 loc) · 2.61 KB
/
metadata-CU-AGE-ST.txt
1
2
3
4
5
6
7
8
team_name: Columbia University
model_name: AGE_STRATIFIED_MODEL
model_abbr: CU_AGE_ST
model_version: "model_03_22"
model_contributors: Marta Galanti (CU) <[email protected]>,Teresa Yamana (CU) <[email protected]>, Sen Pei (CU) <[email protected]>, Jeffrey Shaman (CU) <[email protected]>
website_url: https://github.com/shaman-lab
license: -
methods: Combination of 1) A metapopulation not age-stratified model for a single strain and 2) a state level compartmental model age-stratified and with two strains. The metapopulation county-level SEIR model reproduces transmission within and between the 3142 counties in US. A model-filter framework assimilates daily county incidence of cases and updates the distribution of parameters and state variables at each time step. The inferred parameters and initial conditions are aggregated at the state level and combined with state-level information on population structure, age and state-specific seroprevalence estimates and published age-specific reporting, hospitalization and death rates. Then, to produce age-stratified projections, a second dynamical model, age and population- stratified, is initialized with the previously defined parameters and initial conditions. The age and population-stratified model is a two strains SEIR (Susceptible-Exposed-Infected-Recovered) compartmental model run in isolation for each of the 50 US states and DC. Here the population is stratified in 12 groups in order to reproduce different patterns of disease severity and vaccine prioritization. Specifically, in each state, the population is stratified by years of age (0 - 4, 5 - 17, 18 – 49, 50 - 64 and >65), adult exposure status (essential workers, healthcare workers, other adults) and health risk status (presence or absence of one or more health risk factors for severe disease). The model has a spring/summer forcing on Rt (-2% weekly from march 13th for 3 months) and a fall/winter forcing on Rt (+2% weekly from Sept 15th for 3 months). Data inputs: JHU CSSE County-level confirmed cases, Safegraph mobility, CDC SEROPREVALENCE data from Nationwide Commercial Laboratory Seroprevalence Survey (state, age groups), age distribution of recent infections from individual state departments (state, age groups), POLYMOD contact rates across age group in US. The model does not distinguish between natural and vaccine induced immunity and at t0 all susceptible individuals have some pre-existing immunity (either from vaccine or natural infection). Fore complete methods, see https://www.medrxiv.org/content/10.1101/2020.05.04.20090670v2, https://www.medrxiv.org/content/10.1101/2021.02.22.21252240v1.