params

params is a DataFrame that contains all parameters that quantify how the disease spreads in the "value" column. It can also contain parameters for the contact models such as the degree of assortativeness by a certain variable. Putting those parameters into a DataFrame, allows us to optimize over them using estimagic. Make sure to read about the basic structure of params DataFrames in estimagic, before you continue.

params has a three level index. The first level is “category”, the second is the “subcategory”, the third is called “name”. The values are stored in the “value” column.

We provide epidemiological estimates for many of these variables in the covid_epi_params.csv with explanatory notes and links to their sources which are accessible via

params = sid.load_epidemiological_parameters()

Currently, we have the following categories:

Assortative Matching (assortative_matching)

As the assortative matching parameters depend on the contact models, we don’t provide any defaults. They must be added by the user.

We suggest to implement assortative matching by age_group and region. However, you are free to implement assortative matching by any variable in your states dataset. Having assortative matching not only adds realism to your model but also reduces running time.

For more information on assortative matching see "assort_by".

To tutorials and explanations

Have a look at the simulation tutorial and the assortative matching notebook to see some example contact models and assortative matching parameters.

Health System (health_system)

The default parameters in this category only include the number of free beds in intensive care units which determine how many individuals with serious infection cases survive.

Infection Probabilities (infection_prob)

As the infection probabilities depend on the contact models, wo don’t provide any defaults. They must be added by the user.

To tutorials and explanations

Have a look at the simulation tutorial and the assortative matching notebook to see some example contact models.

Countdowns

Every countdown described in Evolution of States has its own category, describing its distribution.

If the distribution does not depend on the age group, the subcategory is “all”. If the distribution depends on the age group then the subcategory takes the values of the age groups. The states DataFrame then must contain a column called “age_group” with the age groups and their values must match the ones in the subcategory column. In each case the “name” column contains the possible realizations and the “value” column contains the probability. Probabilities for each group must add up to one.

Here is an example with hypothetical numbers:

Hypotetical Parameter Values

category	subcategory	name	value
cd_symptoms_true	all	-1 (= never)	0.25
cd_symptoms_true	all	3	0.75
…	…	…	…
cd_infectious_true	0-9 (age group)	3 (possible value)	0.6 (probability)
cd_infectious_true	0-9 (age group)	5 (possible value)	0.3 (probability)
cd_infectious_true	0-9 (age group)	7 (possible value)	0.1 (probability)
cd_infectious_true	10-20	3 (possible value)	0.6 (probability)
…	…	…	…

The following section describes the epidemiological parameters we provide for Covid-19 and their sources.