This repository follows the guidelines and standards outlined by the hubverse, which provides a set of data formats and open source tools for modeling hubs.
The Consortium of Infectious Disease Modeling Hubs. The hubverse: open tools for collaborative modeling. GitHub release v2.0.0, 14 Jul 2023. URL: https://github.com/Infectious-Disease-Modeling-Hubs
This repository is intended for the Scenario Modeling Hub's research rounds, which are intended to encourage modeling to address specific research questions. Possibilities that have been discussed include the impact of disparities (described below) revisiting early stages of a pandemic. Additional research rounds may follow.
The research rounds are distinct from the operational rounds were the sole focus of the Scenario Modeling Hub from late 2020-April 2024. For information on those efforts, see the operational round GitHub site.
The Scenario Modeling Hub is open to any team willing to provide projections at the right temporal and spatial scales, with minimal gatekeeping. We only require that participating teams share point estimates and uncertainty bounds, along with a short model description and answers to a list of key questions about design. A major output of the projection hub is ensemble estimates of epidemic outcomes (e.g., infection, hospitalizations, and deaths), for different time points, intervention scenarios, and US jurisdictions.
Those interested to participate, please read the README file and email us at [email protected]
Model projections should be submitted via pull request to the model-output/ folder and associated metadata should be submitted at the same time to the model-metadata/ folder of this GitHub repository. Technical instructions for submission and required file formats can be found here, here, for the metadata file, and here, for the abstract file
We observed disparities in COVID-19 outcomes by sociodemographic factors such as age, location, race/ethnicity, occupation, and socioeconomic status, yet most epidemiological models do not account for structural inequities that contribute to differential transmission and severity risk among these groups. This round aims to build multi-model capacity within the Scenario Modeling Hub to model and project how disease is distributed differentially among racial/ethnic subpopulations. A secondary goal is to retrospectively assess the COVID-19 disease burden that could have been averted if various sources of health inequities were reduced or mitigated. The round will focus on two US states, California and North Carolina, where more detailed epidemiological data are available.
We have specified a set of scenarios and target outcomes to allow alignment of model projections for collective insights. Scenarios have been designed in consultation with academic modeling teams and government agencies (e.g., CDC).
In Phase 1, teams will calibrate to case and death data by race/ethnicity from 5/1/2020 – 11/14/2020 and project forward to 11/15/2020 – 4/3/2021 in a single Scenario A. Teams are required to incorporate health inequities that contribute to differential transmission risk and severity by race/ethnicity, where severity is defined as the probability of death given infection. Teams will be evaluated on their ability to model race/ethnicity-specific death time series throughout the projection period in Phase 1. We will assume that we have prescribed the perfect scenario conditions; thus full information about new variant circulation, non-pharmaceutical interventions (NPIs) such as masking, social distancing, and travel restrictions, mobility, and vaccine coverage is known over the entire projection period. As a result, we will be able to test the ability of models to project disparities over time, after controlling for other epidemiological and behavioral uncertainties. Phase 1 features a single scenario as follows:
The goals of Phase 2 are to explore the potential impact of reducing inequities in transmission risk, severity, or both during the same phase of the pandemic projected in Phase 1. In Phase 2, teams will be permitted to calibrate from 5/1/2020 – 4/3/2021 to generate projections for Scenario A (representing empirically observed disparities). This will ensure that teams have captured disparities accurately throughout the entire time period so that the impact of reducing sources of inequities can be soundly assessed in putative Scenarios B-D.
Thus Scenario A can be interpreted as a fitting exercise as opposed to projection. In Scenario B, we assume that all racial/ethnic populations were able to mitigate their risk of non-household transmission to the degree of White populations. This could have been hypothetically achieved through increasing personal protective equipment access and social distancing among essential workers and by increased propensity for telework across occupational industries. In Scenario C, we assume that all racial/ethnic populations experience age-adjusted severity rates upon infection in line with the White population. This should be viewed as a hypothetical exercise in exploring the extent of disease burden that could be averted if long-term infrastructural changes were made to increase quality healthcare access, debias healthcare settings to address institutional racism, and expand institutional/government support systems such as paid leave. In Scenario D, we assume that all racial/ethnic populations experience non-household transmission risk and age-adjusted severity in line with the White population. See the round structure below.
In this retrospective round, we are interested in assessing the ability of modeling teams to capture the racial and ethnic distribution of deaths in California and North Carolina. Therefore, we will provide information on external circumstances that impacted disease dynamics throughout the projection period, including new variants, state-specific nonpharmaceutical interventions (NPIs), mobility, and the ramp up of vaccination during the projection period November 2020-April 2021.
- The Alpha variant B.1.1.7 emerged and replaced the Wuhan strain during the projection period. Alpha B.1.1.7 became the dominant strain mid-March, and was reported to be ~40-60% more transmissible than the Wuhan strain with similar severity (Sah et al. 2021, Yang et al. 2022, Ahmad et al. 2022, Lin et al. 2021). By mid-April 2020, B.1.1.7 represented 66% of all sequenced cases in the United States.
-
- Nov 16, 2020: Mask wearing required in all settings outside of the household and non-essential businesses closed.
- Nov 21, 2020: Curfews from 10pm-5am applied to all non-essential businesses and households.
- Dec 3, 2020: Regional stay at home orders instituted.
- Jan 25, 2021: Regional stay at home orders and curfew are lifted.
-
- Nov 13, 2020: Indoor gatherings limited to 10 individuals.
- Nov 23, 2020: Masking required in non-household settings.
- December 11, 2020: Curfew instated from 10pm-5am.
- Feb 28, 2021: Curfew lifted, masking and social distancing still required.
- March 26, 2021: Restrictions on social distancing are lifted, masking remains in place.
-
Schools were largely closed for the 2020-21 school year in both locations.
-
Weekly SageGraph mobility data is provided at the census tract level, in the disparities/mobility/ folder of covid19-smh-research_resources GitHub Repositories. Mobility data is sourced from Kang et al. 2021.
-
Weekly cumulative age- and race/ethnicity-specific vaccination coverage is provided at the state level in the disparities/vaccination/ folder of covid19-smh-research_resources. Vaccination data is extracted from the CA DPH and NCDPH vaccine dashboards and report vaccination by age and race/ethnicity in separate time series. These dashboards report the number of partially (receiving 1 dose) and fully vaccinated individuals. We describe the details of the vaccination rollout and effectiveness assumptions.
-
We summarize race/ethnicity vaccination data by state into the follow sub-populations: - California: Latino, White, Asian, Black, Other, Unknown - North Carolina: White, Asian, Black, Other, Unknown. Note here that non-White Hispanic individuals are included in “Other.”
-
Vaccination rollout timelines
-
California (COVID-19 Vaccination Plan, CADPH, California: State-by-State COVID-19 Guidance, Husch Blackwell, Vaccines for People with High-Risk Medical Conditions or Disabilities, CADPH)
- December 15, 2020: Vaccination begins for Phase 1A. Healthcare workers and long-term care residents are eligible.
- January 13, 2021: Individuals 65+ years old are eligible for vaccination.
- February 3, 2021: Phase 1B: Essential workers with high exposure risk (those working in agriculture and food, education and childcare, and emergency services) are eligible for vaccination. Vaccination sites are also set up in Oakland and Los Angeles to prioritize communities that have been heavily impacted by the COVID-19 pandemic.
- March 4, 2021: The state announces it will direct 40% of its vaccine supply to vulnerable communities, based on the Healthy Places Index (HPI).
- March 15, 2021: Individuals <65 and >16 years with high-risk conditions are eligible for vaccination.
-
North Carolina (North Carolina: State-by-State COVID-19 Guidance, Husch Blackwell)
- January 20, 2021: Group 1: North Carolinians over the age of 65 and all health care workers who have in-person contact with patients are eligible for vaccination.
- Feb 9, 2020: Group 2: North Carolina mandates a subset of vaccines must go to every geographic region and prioritizes vulnerable communities.
- Feb 25, 2021: Teachers and other Group 3 essential workers can get vaccinated.
- March 3, 2021: Additional essential workers in Group 3 can get vaccinated.
- March 24, 2021: People at higher risk from COVID-19 due to underlying medical conditions will become eligible to receive a vaccine, as will people in certain congregate-living settings (dorms).
- March 31, 2021: Group 4 eligible for vaccination. Essential workers now include frontline workers who do not have to be in person for work and those in a range of sectors such as construction, energy, financial services, public works, and others as categorized by the Cybersecurity and Infrastructure Security Agency.
-
-
VE assumptions: Three vaccines were authorized in the United States: (1) BNT162b2 mRNA from Pfizer/BioNTech; (2) mRNA-1273 from Moderna; and (3) Ad26.COV2.S from Janssen. The efficacy of vaccines against symptomatic infection was 95%, 94%, and 66% for the BNT162b2, mRNA-1273, and Ad26.COV2.S vaccines respectively against the Wuhan and B.1.1.7 strains, with even greater efficacy against hospitalization and death ( Lindsey R. Baden et al. 2020, Fernando P. Polack et al. 2020, Jerald Sadoff et al. 2021, Sara E. Oliver et al. 2020). Ad26.COV2.S from Janssen was not administered in the US until Feb 27, 2021, so it can be assumed it had minimal impact during this study period (Janssen COVID-19 Vaccine FDA Approval Status).
-
Teams are free to ignore the impact of waning given that most individuals would be vaccinated within 4 months of the projection end date. We can assume that VE is 95% after two doses, 50% after one dose, and that doses typically occur 3.5 weeks apart. Teams would maintain these VE estimates for the duration of the study period.
Weekly case and death data by race/ethnicity are available for phase 1 and 2 in the target-data folder. For more information, please consult the documentation associated with the disparities round target data.
In this round, the required target for trajectories will be weekly incident infections, cases, and deaths in California and North Carolina for a set of specified racial/ethnic groups. Trajectories will need to be paired across racial/ethnic groups and horizon (i.e., for a given model, location, scenario and horizon, all race/ethnicity data for simulation 1 corresponds to the sum of race/ethnicity-specific estimates for simulation 1).
In California, required racial/ethnic groups are "latino", "black",
"white", "asian", "other", and "overall".
In North Carolina, required racial/ethnic groups are "white", "black", "asian",
"other", and "overall".
The definitions of race/ethnicity can differ across various datasets.
Given the missingness in demographic disease data and limited data available on case reporting rates by race/ethnicity, it will be optional for teams to submit cases. Infections and deaths will be required and only weekly death targets will be evaluated.
For more information please consult the target data README.
Auxiliary data and code are available in the covid19-smh-research_resources GitHub repository, disparities folder
The folder contains multiple sub-folders:
- Case Imputation: imputed case time series by race/ethnicity at the state level to infer missing case demographic information
- Vaccination: vaccination time series by age and race/ethnicity
- Serology: monthly infection-based and combined vaccination/infection seroprevalence time series
- Population Data: state-level population structure data by age and race/ethnicity
- Hospitalization: hospitalization time series by race/ethnicity, in a rate per 100,000 people for California and number of hospitalizations for North Carolina
- Contacts: synthetic daily contact matrices by race/ethnicity in the household, school, community, workplace setting in the pre-pandemic and pandemic period
| Phase | Type | Scenario | Scenario name | Scenario ID for submission file (scenario_id) |
|---|---|---|---|---|
| 1 | Projection | Scenario A. Inequity-driven transmission and severity by race/ethnicity | phase_one | A-2020-05-01 |
| 2 | Calibration | Scenario A. Inequity-driven transmission and severity by race/ethnicity | inTran_inSev | A-2020-11-15 |
| 2 | Projection | Scenario B. Inequity-mitigated transmission and inequity-driven severity by race/ethnicity | mitTran_inSev | B-2020-11-15 |
| 2 | Projection | Scenario C. Inequity-driven transmission and inequity-mitigated severity and by race/ethnicity | inTran_mitSev | C-2020-11-15 |
| 2 | Projection | Scenario D. Inequity-mitigated transmission and inequity-mitigated severity by race/ethnicity | mitTran_mitSev | D-2020-11-15 |
- End date for fitting data for Phase 1: Saturday November 14, 2020
- Start date for scenarios: Sunday November 15, 2020 (first date of simulated transmission/outcomes)
- Simulation end date: April 3, 2021 (20-week horizon)
- Mid-May 2024 (soft deadline)
- Phase 1 projections due: 2024-06-25
-
Must consist of a subset of weekly targets from Sunday, November 15, 2020 - Saturday, April 3, 2021 (20 week projection period). Weeks follow epi-weeks (Sun-Sat) dated by the last day of the week.
-
Weekly targets: Weekly incident infections, deaths by location and major racial/ethnic group. We require the following racial/ethnic groups by state:
- California:
"latino","black","white","asian","other", and"overall". - North Carolina:
"black","white","asian","other", and"overall".
- California:
-
Optional target: weekly incident cases by location and major racial/ethnic group (as previously stated)
-
We required 100-300 individual trajectories for each target. Trajectories should be sampled in such a way that they will be most likely to produce the uncertainty of the simulated process. Projection quantiles are optional.
- For teams who wish to submit quantiles, the format is in accordance with prior SMh rounds. We ask for the following quantiles: 0.01, 0.025, 0.05, every 5% to 0.95, 0.975, and 0.99
-
Metadata: We will require a brief meta-data from all teams.
-
Filename: the date YYYY-MM-DD should correspond to the due date. For phase 1: the date will be "2024-05-15". For more information, please consult the model-output/README
The repository contains a target-data folder relevant to the modeling efforts. It contains:
- COVID-19 case and death time series by race/ethnicity for the disparities round.
For more information, please consult the associated README file
An additional repository containing the auxiliary data is available on GitHub: covid19-smh-research_resources. This repository contains additional code and data that are relevant to the modeling efforts.
All source code that is specific to the overall project is available under an open-source MIT license. We note that this license does NOT cover model code from the various teams, model scenario data (available under specified licenses as described above) and auxiliary data.
Those teams interested in accessing additional computational power should contact Katriona Shea at [email protected]. Additional resources might be available from the MIDAS Coordination Center - please contact [email protected] for information.
- Shaun Truelove, Johns Hopkins University
- Cécile Viboud, NIH Fogarty
- Justin Lessler, University of North Carolina
- Sara Loo, Johns Hopkins University
- Lucie Contamin, University of Pittsburgh
- Emily Howerton, Penn State University
- Claire Smith, Johns Hopkins University
- Harry Hochheiser, University of Pittsburgh
- Katriona Shea, Penn State University
- Michael Runge, USGS
- Erica Carcelen, John Hopkins University
- Sung-mok Jung, University of North Carolina
- Jessi Espino, University of Pittsburgh
- John Levander, University of Pittsburgh
- Samantha Bents, NIH Fogarty
- Katie Yan, Penn State University