Modelling the spread of Covid-19 in the UK

June Dalziel Almeida

JUNE

Christoph Becker, Richard Bower, Joseph Bullock*, Tristan Caulfield, Carolina Cuesta-Lazaro*,  Aoife Curran, Edward Elliott, Kevin Fong, Richard Hayes, Miguel Icaza-Lizaola, Frank Krauss, James Nightingale, Arnau Quera-Bofarull*, Aidan Sedgewick, Henry Truong, Ian Vernon*, and Julian Willams

 

* speakers

  1. Creating a digital twin of the UK (Arnau QB)
  2. Dynamics of the world (Joseph B)
  3. Modelling Covid-19 (Carol CL)
  4. Policies (Joseph B)
  5. Model fitting and results (Ian V)
  6. Visualizer demo (Carol CL)

Outline

Creating a digital twin of the UK

Setup

Geography

Demography

Population

Venues

  • Households
  • Companies
  • Schools
  • etc.

June's Geography

local authority districts

super area (MSOA) 

area (OA)

North East => 26,000 output areas

June's Demography

male

female

male

female

Demography

What defines a person in June?

  • age  (27)
  • sex (f)
  • ethnic group (Caribbean)
  • deprivation index 2 (1-10)
  • work sector / subsector (healthcare/doctor)
  • mode of transport (public)
  • area of residence
  • super area of work

Main data source: census data (NOMIS)

Where can people be?

  • Residence
    • Care Home
    • Household
  • Primary activity
    • Company
    • Hospital
    • School
    • Care Home
    • University
  • Travel
    • Commute
    • National travel
  • Leisure
    • Shopping
    • Pubs / restaurants
    • Cinema

Residence - Household

  • Households based on census data.

area population

  • number of households that contain at least one person over 65 years old.
  • number of couples
  • ....

16 different household compositions

Residence - Household

Tests

North East

London

Residence - Household

Tests

Housemates matrix (London)

Probability

Residence - Care Home

  • Each area can have up to one care home
  • Location based on census data
  • Three populations: residents, workers, and visitors

What can people do?

  • Residence
    • Care Home
    • Household
  • Primary activity
    • Company
    • Hospital
    • School
    • Care Home
    • University
  • Travel
    • Commute
    • National travel
  • Leisure
    • Shopping
    • Pubs / restaurants
    • Cinema

Primary activity - School

Data on every school in England & Wales

  • Age range
  • Location (coordinates)
  • Number of pupils

Assign each kid to the closest school that fits their age range.

  • Kids are subdivided into year groups
  • We add teachers assuming constant ratio.

Primary activity - Workplace

  • Idea: match residence area => work super area

Flow data

Durham, works in  sector X

Newcastle, industry sector X

Sector info also used for medics and teachers.

Workplace - Residence

City of London workers' usual residence

Workplace - Companies

Company size distribution taken from census data

London

North East

Hospitals

  • We model acute hospital trusts

North East region trusts.

beds

ICU beds

University

  • All universities are included at their exact location.
  • We match communal and student households to university enrolment.

Durham University student households

University

University of Bath student households

What can people do?

  • Residence
    • Care Home
    • Household
  • Primary activity
    • Company
    • Hospital
    • School
    • Care Home
    • University
  • Travel
    • Commute
    • National travel
  • Leisure
    • Shopping
    • Pubs / restaurants
    • Cinema

What can people do?

  • Residence
    • Care Home
    • Household
  • Primary activity
    • Company
    • Hospital
    • School
    • Care Home
    • University
  • Travel
    • Commute
    • National travel
  • Leisure
    • Shopping
    • Pubs / restaurants
    • Cinema

Leisure

  • Exact locations of every single pub, restaurant, grocery store, cinema, etc. from OpenStreetMap

World dynamics

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

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

Modelling Covid-19

Infection requires proximity: social mixing patterns

43 yo

38 yo

10 yo

Contact matrices

43 yo

38 yo

10 yo

43 yo

38 yo

Household contact matrix

BBC pandemic survey: medRxiv 2020.02.16.20023754

Household contact matrix

BBC pandemic survey: medRxiv 2020.02.16.20023754

JUNE

BBC Pandemic Survey

School contact matrix

BBC pandemic survey: medRxiv 2020.02.16.20023754

BBC Pandemic Survey

JUNE

43 yo

38 yo

10 yo

How do people transmit the infection?

j

j

i

Intensity of contacts (per group)

Infectiousness profile

Infectiousness profile

10.1038/s41591-020-0869-5

The fate of the infected

Credit: http://www.eoht.info/page/god+does+not+play+dice

Asymptomatic

Mild symptoms

Severe symptoms

Hospitalised

Intensive care

But it isn't a fair dice!

But it isn't a fair dice!

The symptom trajectories

Policies

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

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

JUNE - IDAS

By arnauqb

Source: carol cuesta

JUNE - IDAS

  • 692

More from arnauqb