Visiting

Household visits

Care home visits

Leisure activities

We include:

Grocery stores
Pubs
Cinemas

Visits and leisure activities

Probability of attending is based on data

p_{\text{attending}} = 1 - p_{\text{not-attending}}\\ p_{\text{not-attending}} = \text{exp}[-\Delta t* \lambda]

Poisson distribution for decision

male_age_probabilities:
  3-75: 0.03
  75-80: 0.01
female_age_probabilities:
  3-75: 0.03
  75-80: 0.01
neighbours_to_consider: 5
maximum_distance: 15
weekend_boost: 1
drags_household_probability: 0.5

Commuting

Commuting offers further mixing potential outside your immediate social and work group

Data on commuting derived from NOMIS 'flow data'

Public

Private

Commuting

11 major stations identified for public commuting

Inter-regional travel

Works in a similar way as commuting

Links created between 11 major cities

Data from National Rail and Department of Transport for departure/arrival numbers

Construct simulation for approximating origin-destination matrix

Validate against data

Free time survey:

Survey of how individuals spend their time during a typical work day

Policies

Policies can be turned on and off according to dates

The following policies have been implemented by the UK Government and are in the model:

Policy	Date
Case isolation at home	12/03/2020
Voluntary household quarantine	16/03/2020
Voluntary working from home	16/03/2020
Voluntary avoidance of leisure venues	16/03/2020
Social distancing	16/03/2020 then 23/03/2020 ('stay at home')
Shielding of vulnerable population	16/03/2020
Closure of schools and universities	20/30/2020
Closure of leisure venues	21/03/2020

Case isolation, shielding and quarantine

Case isolation is the default policy in the model

Implementation:

Anyone with symptoms of COVID-19 must stay at home
Anyone over a certain age must stay at home
People staying at home can still infect other members of the household

shielding:
        min_age: 70
        complacency: 0.7
        start_time: 2020-03-16 
        end_time: 2020-07-04 

quarantine:
        n_days: 7 # for the symptomatic person
        n_days_household: 14 # for the housemates
        household_complacency: 0.6
        start_time: 2020-03-16 
        end_time: 2021-07-04

Leisure avoidance and closure

Implementation:

Leisure venues can be closed or avoided by type

close_leisure_venue:
        start_time: 2020-03-21
        end_time: 2020-07-04
        venues_to_close: ['pub', 'cinema', 'care_home_visit']

change_leisure_probability:
        start_time: 2020-03-21
        end_time: 2020-07-04
        leisure_activities_probabilities:
          household_visits:
            men: 
              0-64: 0.50
              65-100: 0.25
            women: 
              0-64: 0.50
              65-100: 0.25
        leisure_activities_probabilities:
          care_home_visits:
            men:
              0-100: 0.05
            women:
              0-100: 0.05

School and university closure

Implementation:

Schools can be closed by year group
Universities fully close

close_schools:
        start_time: 2020-03-20
        end_time: 2020-07-04
        years_to_close: all
        full_closure: False

close_universities:
        start_time: 2020-03-20
        end_time: 2020-10-01

Company closure

Implementation:

Identify all SIC codes of companies closed and assigned workers as 'furloughed'
Identify all SIC codes of companies containing key workers and assign workers as 'key'
All remaining workers assigned 'random'

Social distancing

Implementation:

Beta reduction by group

social_distancing:
        start_time: 2020-03-16
        end_time: 2021-07-04 # currently unknown
        beta_factor:
                box: 0.5
                pub: 0.5
                grocery: 0.5
                cinema: 0.5
                commute_unit: 0.5
                commute_city_unit: 0.5
                hospital: 0.5
                care_home: 0.5
                company: 0.5
                school: 0.5
                household: 1.0
                university: 0.5

"The strength of association was larger with increasing distance (2.02 change in RR per m)"

Chu et al., Lancet (2020)

Examples

deck

By Joseph Bullock