Australian COVID-19 scenario model

This is an implementation of the disease transmission model described in the paper Coronavirus Disease Model to Inform Transmission Reducing Measures and Health System Preparedness, Australia (doi:10.3201/eid2612.202530).

The following scenarios are predefined by episcen.covid.scenarios():

“Unmitigated” (COVID_AU_Baseline)

Epidemics produced in the absence of interventions.

“Q + I” (COVID_AU_Full)

Epidemics produced subject to case-targeted interventions (quarantine and isolation).

“Dist 25%” (COVID_AU_Dist1)

Epidemics produced subject to case-targeted interventions (as above) and social distancing measures that reduce transmission by 25%.

“Dist 33%” (COVID_AU_Dist2)

Epidemics produced subject to case-targeted interventions (as above) and social distancing measures that reduce transmission by 33%.

_images/COVID_AU_attack_rates.png

Attack rates obtained from each of the four Australian COVID-19 scenarios. Points show median outcomes, and lines show the 5%–95% percentiles.

You can run the scenario simulations and produce the above plot with the following command:

python3 -m episcen.covid

Main functions

episcen.covid.scenarios()

Return the COVID-19 scenario definitions.

Return type:

io.StringIO

Examples:

>>> import pypfilt
>>> import episcen.covid
>>> instances = list(pypfilt.load_instances(episcen.covid.scenarios()))
>>> print(len(instances))
4
>>> ids = set(instance.scenario_id for instance in instances)
>>> print(ids == {'COVID_AU_Baseline', 'COVID_AU_Full',
...               'COVID_AU_Dist1', 'COVID_AU_Dist2'})
True
episcen.covid.plot_attack_rates(png_file='COVID_AU_attack_rates.png')

Plot the attack rates for each scenario.

Note

You must run the scenario simulations before calling this function.

Parameters:

png_file – The PNG file to which the plot will be saved.

Raises:

FileNotFoundError – if the scenario output files do not exist.

Simulation components

class episcen.covid.Model(*args: Any, **kwargs: Any)

The Australian COVID-19 scenario model.

The population is divided into \(N_S\) strata. The state vector contains the following fields:

  • Model parameters (floats):

    • “eta” (scalar): The proportion of infections that present with severe symptoms.

    • “alpha_m” (scalar): The proportion of infections that present with mild symptoms.

    • “R0” (scalar): The basic reproduction number.

    • “R0_scale” (scalar): The scaling factor for R0, which accounts for the transmission probability and contact rate matrices.

  • Infection compartments (integers):

    • “S” \((N_S)\): The susceptible population.

    • “E1” \((N_S)\): The early incubation period.

    • “E2” \((N_S)\): The late incubation period.

    • “I1” \((N_S)\): The early infectious period.

    • “I2” \((N_S)\): The late infectious period.

    • “R” \((N_S)\): The recovered population.

    • “M” \((N_S)\): Managed cases, ascertained upon leaving “I1”.

    • “RM” \((N_S)\): The recovered population of managed cases.

  • Contact-tracing compartments (floats):

    • “CT_m” \((N_S)\)

    • “CT_nm” \((N_S)\)

  • Quarantine compartments (integers):

    • “E1q” \((N_S)\): The early incubation period (quarantined).

    • “E2q” \((N_S)\): The late incubation period (quarantined).

    • “I1q” \((N_S)\): The early infectious period (quarantined).

    • “I2q” \((N_S)\): The late infectious period (quarantined).

    • “Rq” \((N_S)\): The recovered population (quarantined).

    • “Mq” \((N_S)\): Managed cases (quarantined).

    • “RMq” \((N_S)\): Recovered managed cases (quarantined).

  • Event counters (integers):

    • “Cum_Out_I1” \((N_S)\): The number of people who have left “I1”.

    • “Cum_Out_I1q” \((N_S)\): The number of people who have left “I1q”.

    • “Cum_Managed” \((N_S)\): The number of people who have entered “M” or “Mq”.

    • “Cum_Mild” \((N_S)\): The number of people who left “I1” or “I1q”, and have mild symptoms.

    • “Cum_Severe” \((N_S)\): The number of people who left “I1” or “I1q”, and require hospitalisation.

    • “Cum_ICU” \((N_S)\): The number of people who left “I1” or “I1q”, and require ICU admission.

class episcen.covid.AttackRates(*args: Any, **kwargs: Any)

A summary table that calculates the attack rate for each counter in the state vector (i.e., for fields whose name begins with "Cum_").

class episcen.covid.DailyCases(*args: Any, **kwargs: Any)

A summary table that calculates the daily increase for each counter in the state vector (i.e., for fields whose name begins with "Cum_").

Internal functions

episcen.covid.dependent_dists(indep_values, dep_params)

The dependent distributions function for the COVID-19 scenario model.

episcen.covid.dist_alpha_m(eta)

Construct the prior distribution for \(\alpha_m\).

Parameters:

eta – The case-severity parameter \(\eta\).

episcen.covid.calc_contact_matrix(pi)

Return the contact matrix for a stratified population.

Parameters:

pi – An array that defines the fraction of the population that belongs to each stratum.

episcen.util.outflow(n, p, rnd)

Sample the number of individuals that leave each compartment.

Parameters:

  • n (numpy.ndarray) – The number of people in each compartment.

  • p (numpy.ndarray) – The probability that an individual will leave each compartment.

  • rnd (numpy.random.Generator) – A source of random numbers.

Return type:

numpy.ndarray

episcen.util.split_outflow(n, p, rnd)

Split the individuals that leave each compartment into two flows.

Parameters:

  • n (numpy.ndarray) – The number of people leaving each compartment.

  • p (numpy.ndarray) – The probability that an individual will be selected for the first flow out of each compartment.

  • rnd (numpy.random.Generator) – A source of random numbers.

Return type:

(numpy.ndarray, numpy.ndarray)