The Rockefeller Foundation Launches COVID-19 Modeling Accelerator
Fast-Tracks Medical Research into Real World Decision-Making Tools
NEW YORK | April 14, 2021 (updated July 15, 2021) – The Rockefeller Foundation announced the creation of the COVID-19 Modeling Accelerator and the first cohort of projects. This new initiative, which is designed to enable world-class academic research, is expanding access to the latest data, critical decision-making tools, and actionable solutions for state and local policy makers, public health officials, medical care staff, and others working to stop the COVID-19 crisis in the United States.
Right from the start of the pandemic, epidemiologists, and researchers across the U.S. jumped into action, developing novel tools, and providing insights to help state and local governments and communities better respond to the unprecedented crisis. During this time, new tools were launched to calculate the optimal number of times students should be tested for SARS-CoV-2 per week in order to screen for and stop the chain transmission in a schools, as well as to simulate the effects of different lockdown measures and non-pharmaceutical interventions on the number of cases, hospitalizations, and possible deaths that might occur within a specific area, and more.
However, even though some research teams were able to quickly and effectively deliver valuable insights and guidance to local decision makers, many researchers still faced significant challenges associated with translating an increasing amount of academic research into decision support tools. This includes how to align tools with the real needs of decision makers, secure funding to do often unmandated research, create user-oriented software or interfaces, and adapt workflows to include user research and feedback.
Solution: Accelerate Access to Decision-Making Tools
To help research teams overcome these challenges, The Rockefeller Foundation funded the Society for Medical Decision Making (SMDM) which is overseeing the Accelerator. As a result of this unique combination of grant funding and technical assistance, the Accelerator serves as a comprehensive approach to increasing access to, and the development and implementation of, innovative and actionable decision-making tools and resources.
“It became clear early on in the crisis that we needed a new mechanism to quickly translate the incredible amount of critical research being published all across the country into useful, practical tools and solutions,” said Evan Tachovsky, Director and Lead Data Scientist at The Rockefeller Foundation. “The Accelerator pairs funding with world class data product support to help academic teams build algorithms, dashboards, and code bases that are genuinely useful for decision makers.”
To implement this project, SMDM partnered with leading experts from Johns Hopkins University, Duke University, University of Michigan, as well as product design firm Fenris to select, fund, and advise research teams from across the United States. The Accelerator, which has already started to award technical assistance and funding on a rolling basis throughout the crisis, will remain flexible as it promotes access to and availability of vaccines, testing, treatments, and more. Initially, it launched five teams and has recently funded three additional teams.
The teams in this first cohort share the common goal of using modeling techniques to aid policy makers, public health officials, and medical system staff make more informed and effective decisions amid the pandemic. Today, Accelerator teams are already serving decision makers and communities across the U.S., including the University of Michigan’s MI Safe Start, which serves counties and communities across the state, and the COVID-19 Simulator Tool, which helps state and local decision makers model and forecast risk.
COVID-19 Modeling Accelerator Grant Recipients
- COVID-19 Healthcare Projection Dashboard to Inform Local Decision-making throughout the United States
Lauren Ancel Meyers, University of Texas at Austin; Michael Lachmann, Santa Fe Institute
This team at UT-Austin will extend the Texas COVID-19 Dashboard to provide weekly forecasts of COVID-19 hospitalizations and ICU patients for every county in the US using a validated model of COVID-19 transmission dynamics driven by local hospitalization and cellphone mobility data. It will include interactive visualizations of past hospital and ICU census data, one-month ahead healthcare forecasts, and the past and current transmission rate of the virus for all counties, metros, states and the entire US. This tool will facilitate real-time risk assessments by policymakers, public health agencies, schools and the public, and provide vital projections to inform timely countermeasures and ensure sufficient ICU and hospital capacity.
Ryan McGee, Carl Bergstrom, University of Washington
SEIRS+ is an open-source Python framework that supports customizable implementations of extended SEIR models that embrace stochasticity, demographic heterogeneity, and the structure of contact networks. The primary goal of SEIRS+ is to support decision makers and researchers with a set of modeling tools that a) rigorously capture realistic dynamics of transmission and intervention, b) are flexible and customizable, and c) are easy to set up and run. To date, SEIRS+ has been used to study and inform mitigation strategies in workplaces, universities, K-12 schools, and nursing homes. Moving forward, the goals for SEIRS+ development include extending the model to include features of relevance to the COVID-19 pandemic in 2021 (e.g., temporal vaccination dynamics and tracking multiple strains with different characteristics) and furthering the frameworks’ ease of use for a broader set of users and decision makers. These efforts aim to increase decision maker access to robust epidemiological models that are tailored to their particular communities and that enable quantitative assessment of epidemic trajectories and the effects of mitigation strategies on these outcomes.
- MI Safe Start Map
Marisa Eisenberg, Paul Resnick, Sharon Kardia, Kirtana Choragudi, Jessie Singh, Michael Hayashi, Michael Hess, Emily Martin, Josh Petrie, Jon Zelner, University of Michigan
MI Start Map is a dashboard designed to monitor the status of COVID-19 indicators across the state of Michigan. The goals of MI Start Map are twofold: 1) to assist public health officials in making state, regional, and county-level decisions related to COVID-19; 2) to provide the general public with insight into some of the indicators that affect these public health decisions. The MI Start Map umbrella includes MI Lighthouse, a private dashboard for authorized users only, designed to provide state and local public health officials with critical symptom, case, and vaccine data to assist with outbreak investigations and resource allocation decisions. The team will add new layers to both the public and private MI Start Map to track vaccination and health disparities information. Expanding the MI Start Map umbrella in these two directions will allow us to: a) improve understanding of COVID-19-related disparities for the public and decision-makers, and b) enable vaccination rapid response for policymakers and state and local public health.
- COVID-19 Simulator: Simulation Modeling to End the COVID Pandemic
Jagpreet Chhatwal, Massachusetts General Hospital, Harvard; Benjamin Linas, Boston Medical Center; Turgay Ayer, Georgia Tech
This multi-institutional COVID-19 Modeling Consortium will extend their COVID-19 Simulator (www.covid19sim.org), an interactive publicly available online tool, to inform the timing of lifting restrictions and going back to normalcy for each state considering the COVID-19 vaccine uptake and new SARS-CoV-2 variants of concern. The COVID-19 Simulator utilizes a system dynamics (compartment) model to project the trajectory of the COVID-19 pandemic in each state, and its results are used in the weekly COVID-19 forecasts by the Centers for Disease Control and Prevention (CDC). For each selected intervention scenario, the COVID-19 Simulator projects and visualizes the number of deaths from COVID-19; the number of active, diagnosed, and undiagnosed COVID-19 cases; and the number of hospital beds and intensive care unit (ICU) beds needed for COVID-19 patients in each state.
- COVID-19 Fairness in Resource Allocation (FIRA) Engine
Corey Jackson, Michael Ferris, University of Wisconsin—Madison
With the rapidly increasing availability of COVID-19 vaccinations, federal, state, and local decision-makers have turned to computational modeling to guide their decisions about vaccine Management (such as the wiVDM Model in Wisconsin). While these models have helped optimize vaccine delivery and allocations, they have not successfully overcome the existing health disparities that stem from a history of institutional injustices, including unequal access to health care, discrimination, and gaps in education, income, and wealth attainment. The team will build a catalog of fairness interventions and assessing the interventions to determine if they yield positive outcomes (i.e., increasing immunization rates). The inclusion of fairness recommendations will allow decision-makers to determine how interventions could lead to increased immunization during allocation planning. This will enable decision-makers to become aware of actual fairness interventions and the range of potential interventions to improve vaccine management. The team expects that this model will be helpful for the state of Wisconsin and could be useful for other states and geographies and problem spaces where equity is a factor.
- Vaccines Planning Tool
Lauren Ancel Meyers, University of Texas at Austin; David Morton, Northwestern University
During the COVID-19 pandemic, government agencies worldwide developed staged alert systems that monitor data streams and trigger changes in non-pharmaceutical interventions. Most of these systems were developed prior to the rollout of effective SARS-CoV-2 vaccines. With a significant fraction of the US population immunized against the virus, state and local decision makers face a new set of challenges. Coupling SARS-CoV-2 transmission dynamic models with numerical optimization techniques, we are developing tools to support (i) strategically closing gaps in vaccine coverage within US cities, and (ii) the design of staged alert systems that track and manage risks associated with emerging variants in partially vaccinated populations. These tools are based on methods we developed when designing the staged alert system that has governed COVID-19 mitigation policy in the Austin, Texas metropolitan area since May 2020.
- Vaccine Equity Planner
John Brownstein, Ben Rader, Kara Sewalk, Boston Children’s Hospital
Boston Children's Hospital's Computational Epidemiology Lab and Ariadne Labs collaborated to develop the COVID-19 Vaccine Equity Planner to support state and county government and public health decision-makers to visualize "vaccine deserts", identify sites to address needs, and support providers to engage communities. Powered by VaccineFinder data of all active COVID-19 vaccination sites, the tool combines geospatial data, maps and travel times provided by Google, in addition to potential vaccination sites, population characteristics and area-based measures to present vaccine deserts on an interactive map. Using the CDC's Social Vulnerability Index and the Delphi Group's COVID-19 symptom survey, it highlights particularly vulnerable deserts, other potential sites (e.g., primary care practices, pharmacies, schools, places of worship) within the deserts and population characteristics, such as vaccine intent and hesitation. This tool provides real-time updates and the ability to download a list of alternative distribution locations, empowering local public health professionals to address distribution gaps in their own community.
- Hospital Capacity Management through Optimal Patient Transfers
Kimia Ghobadi, Felix Parker, Johns Hopkins University
The team at Johns Hopkins will extend and customize their COVID-19 Hospital capacity management dashboard to find optimal patient transfer strategies for healthcare systems. With the COVID-19 surges, many hospitals face capacity challenges as they balance care for both covid and non-covid patients. This burden can be reduced by transferring patients proactively and optimally from over-capacity hospitals to nearby hospitals with available space. Our user-friendly and interactive decision-support dashboards will feature input data capabilities, on-the-fly mathematical optimization models, and interpretable and actionable insights.
The Rockefeller Foundation advances new frontiers of science, data, and innovation to solve global challenges related to health, food, power, and economic mobility. As a science-driven philanthropy focused on building collaborative relationships with partners and grantees, The Rockefeller Foundation seeks to inspire and foster large-scale human impact that promotes the well-being of humanity throughout the world by identifying and accelerating breakthrough solutions, ideas, and conversations. For more information, sign up for our newsletter at rockefellerfoundation.org and follow us on Twitter @RockefellerFdn.
About the Society of Medical Decision Making
Founded in 1979, the Society for Medical Decision Making is a not-for-profit, professional research organization of approximately 1,000 members worldwide. SMDM promotes scientific and methodological rigor in health care decision research and its application to health policy and clinical care. SMDM is the leading society for studying and advancing decision sciences in health, including incorporation of patients’ values and preferences. As a professional society, SMDM brings together experts from numerous fields, including economics, psychology, sociology, education, communication, mathematics, organizational theory, clinical epidemiology, public health, and clinical medicine. The mission of SMDM, reflecting this interdisciplinary approach, is to improve health outcomes through the advancement of proactive systematic approaches to clinical decision-making and policy formation in health care by providing a scholarly forum that connects and educates researchers, providers, policy makers, and the public. For more information, sign up for the SMDM newsletter at smdm.org and follow on Twitter @socmdm, and the Accelerator leaders: @kathymcdonald, @GSchmidler, @laprosser
Fenris is a consultant network created by Aman Ahuja to support teams developing responsible data systems for public benefit. We seek to make these ventures successful and sustainable through product strategy, data governance, team facilitation, and technical assistance. Learn more about our work with the COVID Modeling Accelerator at amanahuja.me/cma and follow Aman on twitter at @AmanQA.
+1 (646) 465-0885
Dr. Kathy McDonald
Society of Medical Decision Making