Show HospitalsShow hospital locations and resources for chosen state.
Select Region(s) for Comparison
Data to Graph
Select Region and Similarity Type
Data to Graph
Click a row in the data table below to compare NPI actions and timeframes between the selected region and the clicked region.
State Comparison Analysis Tool
Data to Graph (Y Axis)
Data to Graph (X Axis)
Additional Simulation Parameters
The settings below drive the underlying parameters that generate the figures and statistics above.
Health System Capacity
Data to Graph
Data to Plot: Timeframe
Data to Plot: Hotspot Criteria
Data to Graph
COVID-19 Dashboard and Tools
These dashboards and model provide state and county-level insight into the dynamics of COVID-19 regarding hospital capacity, PPE, and non-pharmaceutical interventions (NPIs). This collection of tools developed by the COVID-19 Healthcare Coalition in partnership with the Department of Homeland Security provides access to data and models to understand the secondary effects of key decisions, such as adding hospital capacity or lifting interventions early, and how the fatality rate will change if hospitals become overwhelmed. This tool is built with publicly available data and models.
Use the left-hand navigation bar or click an image below to explore data you care about most.
Questions about this tool can be directed to the owners of each tool listed below.
C19HCC Decision DashboardAn executive decision-aid for policy makers
The Decision Support Dashboard was inspired by the National Governor's Association's Roadmap to Recovery report release in April 2020. This report recommends key metrics and decision criteria necessary to reopen, manage, and react to the on-going pandemic. This Dashboard uses location specific data to help governors, mayors, business owners, and other community-based decision makers answer their pandemic related questions. It simplifies a tremendous amount of health specific data, turning dense information into stoplight harts
Regional DataView the geographical spread of COVID
This view presents a dashboard to discover trends in current cases, deaths, resources, testing, and mobility of a given region. Trends are shown temporally with a time-series chart, and across regions with geo-spatial maps. An interactive table to the right of the map allows you to quickly learn the most at-risk locations with respect to a chosen metric. Data can be viewed at the national, state, and county level; simply mouse over areas to learn more and click on them to zoom in.
Regional Comparison ToolCompare the progression of different areas
Select a country, state, county, or metro area by typing the name of the area into the appropriate text box. This will allow you to filter and then select the appropriate region. You may select as many areas as desired in any combination. As you do so, a graph of deaths will appear on top, and confirmed cases below. Select from the Begin Data drop down to line up the data chronologically, beginning with the first case, first death, first 100 cases, time when exceeded 0.2 cases per 100,000 people, or by the date of a Stay at Home order. It is useful to line up areas in different ways to make visual comparisons of trends.
You can use the Show Data As dropdown to view the data as: raw data, on a logarithmic axis, the 3-day change, the daily growth rate, the doubling rate, or as incidence per 100,000 persons. When the Overlay Region-Specific NPIs is toggled on, each time series will be annotated with a vertical line and code for the data of NPI for that region. The keys for these NPI codes are in the table labeled NPI Key .
Similar Regions ToolIdentify similar regions based on COVID cases and demographics
Select a state and county by typing the name of the area into the text boxes. This will allow you to filter and then select the appropriate region. The tool will automatically generate a list of regions (counties, states, metro areas, and nations) that are similar to the selected region. A graph of confirmed cases will appear to the right. Select a Similarity Type from the dropdown list; definitions of these types are included in the table labeled Types of Similarity . The Begin Data selection is automatically set to From First 20 Cases, with Other Region Alignment which shifts the case curves of the most similar regions up to 1 week in either direction so that they are as closely aligned as possible. You may select other Begin Data options. You can use the Show Data As dropdown to view the data as: raw data, on a logarithmic axis, the 3-day change, the daily growth rate, the doubling rate, or as incidence per 100,000 persons. When the Overlay Region-Specific NPIs is toggled on, each time series will be annotated with a vertical line and code for the data of NPI for that region. The keys for these NPI codes are in the table labeled NPI Key . Regions with fewer than two weeks of at least 20 cases are excluded.
Mortality Analysis ToolInvestigate trends in the virus's mortality rate over time
This view explores if and how the COVID-19 mortality rate is changing over time. As the risk of death due to COVID-19 increases with age, this analysis uses CDC datasets to estimate the distribution of cases and deaths across different age groups over time, then stratifies the reported COVID-19 case and death counts into age cohorts using these distributions. The mortality rate for each age group is then calculated after applying a lag between case deification and death.
To explore this analysis, select the Data to Graph from the following: Mortality Rate by Age Group, Distribution of Cases by Age Group, Distribution of Deaths by Age Group, and Testing Availability.
State Comparison AnalysisCompare key COVID and behavioral metrics across all states on a scatterplot
This tab allows you to visualize understand how changes in State-level policies are affecting trends in COVID-19 case trends and population behavior. We use these scatterplots to compare states according to population mobility, growth of COVID-19 within the state, and percent positive testing rate. From these scatterplots, we can identify clusters, trends, and potential outlier states to better understand any potential relationship between a state’s policy and their epidemic outcomes.
NPI Analysis DashboardCompare the Evolution of NPIs Across States
Another C19HCC dashboard that visualizes the impact of Non-Pharmaceutical Interventions (NPIs) in the United States over time. It allows for comparison of States with respect to the timing of NPI implementation and their impact on the spread of the virus. Unlike our tool, this dashboard is purely data-driven (i.e. no simulation)
Short-Term ForecastShort-term estimate of COVID cases
Select a state and county by typing the name of the area into the text boxes. This will allow you to filter and then select the appropriate region. A graph of the estimated and actual number of cumulative cases will appear to the right. This forecast uses curve-fitting to generate these estimates, and should not be considered reliable for long-term estimates.
Intervention Model"What If" Modeling
This tab uses an SEIR model to understand the dynamics of COVID-19. The Intervention model used in this tool is built using an adaptation of the model created by Alison Hill and licensed under a Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) License . SEIR models are a primary tool in epidemiology, providing insight into epidemiological dynamics, but are not well-suited for accurate near-term forecasts given their dependence on parameters that are difficult to estimate and assumptions about population mixing.
We have attempted to "fit" transmission characteristics to local areas by finding optimized model parameters that best fit the actual data for COVID-19 deaths for a given area; often however, these fits are not satisfactory.
Using Model Interventions
Start by choosing a U.S. state from the top dropdown list. With this, the model is loaded with basic data describing that state, including the population, case history, hospital capacity, and the dates of any NPIs implemented by the state. Each state has different transmission characteristics. After selecting a State to model, select what metrics to graph. The choices are:
- Critical Infections and ICU Bed Capacity
- Critical Infections and Ventilator Capacity
- Cases Needing Hospitalization and Hospital Bed Capacity
- All Infections
Inspect the Graph
The graph to the right plots the modeled value selected under Data to Show . The red line indicates the present day, and the purple shaded region is the time period during which NPIs are in place. If the value of the metric being graphed has a very high maximum value in the shown time period, the early numbers may be small and difficult to see. You can click and drag on a region of the chart to zoom in on that area. When you move the cursor over any data line, a tool tip will appear with the values at that date. More viewing options are available in the top right of the graph, such as zooming, panning, resetting the view, and tooltip style.
Underneath the graph is a timeline of NPIs currently in place for a given state, starting on the actual date each went into effect. As most states have not yet announced when NPIs will be lifted, each NPI is kept in place for the duration of the simulation. The length of these NPIs can be changed by dragging the bars to the desired time, and then clicking on the timeline canvas to commit the change. Finer control on data selection can be achieved through the date picker in the NPI tab, described below.
Overlay Actual Data
Underneath Data to Show is an option to Overlay Actual Data . Click the toggle to overlay actual data when available for deaths and confirmed cases. Actual data for deaths is only overlaid when Deaths is selected as the metric to graph, while confirmed cases are only overlaid when All Infections is selected as the metric. These models were fit to the number of deaths, so the modeled number of all infections includes the many infections that go unconfirmed and may be substantially higher than the number of confirmed cases due to limitations in the availability of testing.
This tab shows the five primary NPIs used. By default, all NPIs currently in place in that state are selected, beginning on the date they were enacted, and held in place for four months. The dates of each NPI can be changed using the date picker next to the name or the timeline canvas described above. As the NPIs change, the shaded region in the graph will also change.
Currently, the models of NPI effectiveness are rough estimates, chosen based upon observations from Google's COVID-19 Mobility Reports SafeGraph's Geographic Response to Shelter in Place , and real world observation. This is an active area of model development, and it is known that different locations have varying levels of effectiveness, and that this effectiveness varies over time. The model does not yet account for this.
The values for NPIs can be adjusted in the Advanced Settings tab if desired. An NPI with a value of 1.0 would mean that it is 100% effective at reducing contact between people; 0.5 would indicate a 50% reduction.
Health System Capacity
Under the NPI tab, actual state data for hospital beds, ICU beds, ventilators, and their average usage is shown. By varying these parameters, e.g., adding more capacity, you can observe the impact this has on the cumulative number of deaths.
This data does not currently include information on local field hospitals that have been deployed in response to COVID-19.
This tab can be found in the left sidebar. In Alison Hill's original model, there are many parameters for the Intervention Model. We have chosen default values based on current data becoming available, with a bias toward U.S. data. Additionally, we have added more parameters for the effectiveness of NPIs, which can vary. All sources can be found in the Sources tab at the left. When fitting the model to individual states, four parameters were varied: population size, the date of the first case, initial # infected, mild transmission rate, severe transmission rate, and critical transmission rate. Please experiment with different values to build insight, especially for highly known parameters, such as the percentage of asymptomatic carriers there are in the population.
Model Comparison ToolCompare COVID projections from other leading state-of-the-art forecasting models.
With this tool, you can compare forecasts from all the top models cited by the CDC as tools to help health policy officials direct responses to COVID-19. Each model has its strengths, limitations, and sets of assumptions, but when viewed together, they can give a more holistic forecast than one model alone.
Hotspot IdentificationView current and forecasted hotspots at the county level
This tool allows you to visualize current COVID hotspots across the country at the county level, and forecast where future hotspots will be based on current trends. By default, hotspots are defined by looking at the change in cases per capita over the entire duration of data, but you can adjust the criteria to look at totals, or shorten the time-frame to 1 or two weeks.
PPE & Pharma ModelForecast demand for PPE and pharmaceuticals at the state and county level
This tool generates short-term predictions of future demand for personal protective equipment (PPE) needed by healthcare and other essential workers when in contact with COVID-19 patients. Model predictions are driven by the confirmed number of COVID-19 cases within a chosen county. In addition, this tool can also predict future demand for pharmaceuticals, such as Aspirin or AspirinDexmedetomidine.
COVID-19 System Dynamics NPI ModelModel the spread of COVID-19 and test the efficacy of interventions
This tool allows you to model the spread of COVID-19 using a highly flexible SEIR model. Unlike many existing models, it allows you to model both pharmaceutical interventions (vaccine effectiveness, quality of COVID treatment) and non-pharmaceutical interventions (improved hygiene, mask wearing, and social distancing). While it is calibrated on state-level case data, it should not be used for forecasts.
About the Model
The COVID-19 Interventions model is a system dynamics model of the Coronavirus pandemic based on a SEIR (Susceptible-Exposed-Infectious-Recovered) epidemiological framework. The susceptible population becomes exposed to the virus through contact with the infectious population. After an incubation period the exposed population becomes symptomatic with varying levels of severity (mild, severe, and critical) and in critical cases it may be fatal. The infectious population recovers and is immune to the virus. Interventions are layered over the model. Non-pharmaceutical interventions (NPIs) such as social distancing, masks, and hygiene reduce contact with the infectious population and reduce the rate of infectivity. Healthcare capacity can improve outcomes for hospitalized patients. Fatalities can be curtailed by sufficient beds and ventilators. Infection among providers can also be reduced with a higher availability of Personal Protective Equipment (PPE).
Questions about this tool can be directed to Dr. Chris Glazner (email@example.com) .
Advanced Settings: Clinical & Simulation Parameters
Disclaimer: The models and projections shown in this application are computed with the most-current and/or widely accepted clinical and population-health related parameters found in the literature and empirical data. Most of these parameters are documented in the SOURCES page, and others were derived via a collaborative process with internal and external Subject Matter Experts, health-policy experts, and clinicians. This being said, there is a high degree of uncertainty surrounding many of these parameters given the scale and speed at which COVID-19 has developed. Therefore, the models and projections found across this application may be recalibrated using the interactive sliders and inputs found on this page. Adjustments made to these parameters should be done with extreme care and policy interventions evaluated using custom parameter values may not be as accurate as the default parameters proposed.
Additionally, parameters that have been analytically and automatically tuned to better align with empirical data are denoted with ( )