Forecast Modeling: Methods

The Idaho State Forecast models the spread of COVID-19 using a compartmental model. The compartments in a model represent different states that disease individuals may pass through. The disease pathology of COVID-19 fits naturally into a model that is of the (S)usceptible–(E)xposed–(I)nfectious–(R)ecovered type; SEIR models have successfully been applied to influenza pandemics in predicting the impact of the emergence of a novel disease variants.

The basic structure of compartmental models consists of the states that individuals may occupy coupled with rates at which individuals switch between states.We include two types of networks within our differential equations. The first is an explicit contact network. The contact network is a critical feature of our because it shapes transmission within populations. The second network is the spatial network that connects populations and determines the spatial spread of the disease through the area of interest; in other words this network shape transmission among populations. In combination, our model is both age-structured and spatially-structured. In the contact network, nodes are age groups (e.g., 0–14 years old) and the edges describe the contacts × person − 1 × day − 1 between nodes. We use the rates determined by to parameterize the default contact network. For nodes in the spatial network, we currently use the largest 50 cities Idaho by population. Edges in the network are given by using inverse-distance-weighting with either the straight-line distance (based on latitude and longitude) or the driving distance between cities.


The model takes data from the Idaho Department of Health and Welfare, the Idaho Division of Public Health Tableau, mobility data from Google, demographic data from the US Census, and estimates on diseases severity from the CDC Morbidity and Mortality Weekly Report.

More questions? Contact us!