Illinois Covid Modeling is a joint effort between Northwestern University (NU), the University of Chicago (UC), the University of Illinois at Urbana-Champaign (UIUC), and Argonne National Laboratory (ANL) who have been modeling COVID-19 in Illinois since March 2020. In partnership with the Illinois Department of Public Health (IDPH), we use simulation models to predict trends in hospital usage and deaths at the COVID-19 Region level. The NU and UC models are SEIR-based compartmental models, the UIUC model is an age-of-infection model, and the ANL model is an agent-based model. The models are calibrated to these data sources collected by IDPH:

  1. daily census of non-ICU hospital beds occupied by COVID-19 patients (NU, UC, UIUC, ANL)
  2. daily census of ICU hospital beds occupied by COVID-19 patients (NU, UC, UIUC, ANL)
  3. COVID-like illness (CLI) hospital admissions (NU, UC)
  4. COVID-attributed death data in hospitals (UIUC)
  5. COVID-attributed death data in and out of hospitals (NU, UC, UIUC, ANL)
The NU, UC, and UIUC teams have created a suite of 11 models, one for each COVID-19 Region. UIUC also created a 12th model for the state as a whole, while Illinois-level predictions for NU and UC are summed up from their individual Region models. The ANL model is currently producing results for COVID-19 Region 11. We continuously perform other analyses of pandemic dynamics, some of which are featured on the Reports page.

Code for Northwestern's simulation model is available here. Northwestern is estimating \(R_t\) by applying the epyestim package to the timeseries of daily new infections inferred from the simulation models.

Code for University of Chicago's simulation model is available here. The University of Chicago group is estimating \(R_t\) using methods in EpiNow2 to hospital admissions with COVID-like illness, which have tracked actual COVID-19 admissions very well. These estimates are checked against the \(R_t\) derived from the infection rates that are inferred from the dynamical model.

Code for UIUC's simulation model is available here and the details of their procedure simultaneously fitting four data streams is here.

The Argonne model (CityCOVID) is a distributed agent-based model built on the free and open source Repast for High-performance Computing (Repast HPC) agent-based modeling toolkit and the Chicago Social Interaction Model (ChiSIM) framework. \(R_t\) is calculated by tracking individual simulated agents across their infectious careers and tallying up the number of secondary infections that they produce. The \(R_t\) value for a date is the average number of secondary infectious produced by all the agents who began their infectious careers on that date.

Case and death data shown on the landing page are aggregated from county-level public data from IDPH .

About the modeling teams

COVID-19 modeling at Northwestern is led by the Malaria and COVID-19 Modeling Team, an infectious disease modeling group with expertise in using models to inform policy. In addition to their COVID-19 work, they work with malaria-endemic countries to develop data-driven national strategic plans to reduce malaria morbidity and mortality.

The Cobey Lab at the University of Chicago is an infectious disease modeling group with expertise in respiratory pathogens. In addition to their COVID-19 work, they study how the host adaptive immune response coevolves with pathogens, especially in ways relevant to epidemiological and evolutionary forecasting, vaccine design, and pathogen diversity.

The UIUC modeling team led by Professors Goldenfeld and Maslov includes scientists from the University of Illinois and Brookhaven National Laboratory. In addition to their COVID-19 work, the leads of the team study dynamics and co-evolution of microbial ecosystems including microbial pathogens as a part of Biocomplexity theme at the Carl R. Woese Institute for Genomic Biology, University of Illinois.

The CityCOVID team at Argonne National Laboratory is led by Charles "Chick" Macal and Jonathan Ozik from the Decision and Infrastructure Sciences Division. The group develops large-scale agent-based models in a variety of domain areas, including healthcare, energy, and the environment, to support decision making. The group also publishes open-source toolkits for agent-based modeling and high-performance computing.