A tale of two contact-tracing apps: comparing Australia's COVIDSafe and New Zealand's NZ - International ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
A tale of two contact-tracing apps:
comparing Australia’s COVIDSafe and New Zealand’s NZ
COVID Tracer
Bronwyn Howell, School of Management, Victoria University of Wellington
Petrus Potgieter, Department of Decision Sciences, University of South Africa
Presented at the International Telecommunications Society webinar,
Tracking and Tracing Apps for Covid-19, September 29 2020
1
Context
• COVID-19 pandemic
• March 2020:
– both countries adopt an “elimination”
strategy:
– economic and social ‘lockdowns’
(stronger in NZ)
• May 2020
– both countries appear to be succeeding
in elimination
– both introduce smartphone-based
contact tracing apps
• in Australia prior to lockdown-
easing
• in NZ after lockdown-easing
• July 2020 – resurgence of community
transmission in Australia (Victoria)
• August 2020 – resurgence of community
transmission in NZ (Auckland)
2
The Apps
3
Australia: COVIDSafe
• Bluetooth-based
– Singapore’s BlueTrace open
source software
– data stored on Amazon Web
Services within Australia
• Users ‘opt in’
– download from Apple or Google
app store
• Cryptographically-encoded
“handshake” exchanged when any
two app-enabled phones
– come within 2 metres of each
other
– and remain in that proximity for
at least 15 mins
• Handshake data deleted after 21 days
4
When Smartphone owner tests COVID-19
+ve
• User decides whether to upload
phone data to Contact Tracing
service
• Contact Tracer
– combines phone, interview
data
– determines contacts to “list”
– automated messages sent to
“listed” “contacts” identified
on central system
• Optional Bluetrace feature
– system can automatically
broadcast a message to all
contacts
5Effects on Contact Tracing Processes
• Risks
– app may identify “false positives”
• phones 2m apart for 15 mins
but physically separated
(e.g. offices, apartments)
• if using automated
messaging, “too many”
people may present for
testing, causing bottlenecks
in this part of the system
– “false negatives”?
• cannot detect virus
transmission from particles
on surfaces
• Summary
– looks promising
6NZ COVID Tracer
• QR code-based “digital diary” kept by
individual app users
– extends from MBIE contact register
obligations for businesses operating
during lockdown
• Businesses register, download and display
code
– unique for each business premise;
mandatory from August
– initially required MBIE business
registration, subsequently relaxed to
allow unregistered (informal)
businesses to participate
• Users download app
– register contact information initially
– then scan codes every time a premise is
entered
• Phone data deleted after 31 days
7When Smartphone owner tests COVID-19
+ve
• Phone user decides whether to upload
phone data to Contact Tracing service
• Contact Tracer
– combines phone data with interview
data
– identifies premises user has visited
when infectious
– requests contact register information
from business
– parses business information to identify
and “list” individuals who may meet the
2m/15 min criteria
• Optional automated alert
– users can opt to be alerted if they have
entered a premise at the same time as
someone testing +ve
8Effects on Contact Tracing Processes (New
Zealand)
• Contact identification
– app identifies locations visited,
not individuals
– separate processes for premise
records
• Premise registers a new feature
– adds a step to contact-tracing
– imperfect records – individual
scans in but not out
• no closeness of proximity or
length of exposure
• risk contingent on premise
characteristics
• requires explicit human
judgement
9Effects on Contact Tracing Processes
• Business records of who else was
present
– highly variable form (including
paper records)
– difficult and/or costly to parse
• Creates more work for contact tracers
– increases the more businesses
participating and the more users
scanning their activities (more
target premises identified)
• So likely will reduce contact tracer
efficiency relative to the non-app
counterfactual
– greater manual effort required
the higher is adoption and use
10Other factors (New Zealand)
Automated messaging
– does not require contact tracer time
– but will generate a large number of “false positives”
– huge increase in numbers presenting for testing
• delaying testing of true positives
• thereby militating against elimination objective
What about contact at non-business premises?
– e.g. parties at private homes (where Bluetooth apps still
operate)
Large costs in addition to contact tracing
– business obligations; “scanning fatigue” amongst users
But does ensure up-to-date contact info is recorded 11The proof of the pudding will be in the eating
Recent
resurgences have • low uptake
been of little help • and in NZ’s case low usage
in evaluation
NZ app now • even though no standardisation
mandatory for of collection and
businesses but
not users • processing of business records
12NZers are not “Dining”
Daily App Downloads and Scans
250 000 3 000 000
2 500 000
200 000
2 000 000
150 000
Apps
Scans
1 500 000
100 000
1 000 000
50 000
500 000
- -
2020-05-20 2020-06-20 2020-07-20 2020-08-20 2020-09-20
Date
App Scans
registrations
13Scans per app per day
1,40
1,20
1,00
0,80
0,60
0,40
0,20
0,00
14QR Code Activity
8,00 400 000
7,00 350 000
6,00 300 000
5,00 250 000
4,00 200 000
3,00 150 000
2,00 100 000
1,00 50 000
0,00 -
Scans per QR Code per day Cum QR Codes
15Integration with human processes
16Context: Contact Tracing Processes
• Key plank of elimination
strategy
• Interrupt chain of virus
transmission
– identify and isolate close
contacts as soon as
possible
• Three steps
– close contact
identification
– contact listing
– contact follow-up
17Standard Contact Tracing Processes
Supplement data
Contact Tracers
Index Identify
ary data
case Contacts
Notify Contact/
contacts Follow up
testing
yes
Close Contacts
yes Prese no yes
Test Tested
nt for exit
+ve? -ve?
test?
no
18COVIDSafe and Contact Tracing Processes
Supplement data
Contact Tracers
Index Identify
ary data
case Contacts
Notify Contact/
contacts Follow up
testing
yes
Close Contacts
yes Prese no yes
Test Tested
nt for exit
+ve? -ve?
test?
no
19Standard Contact Tracing Processes
Supplement data
Contact Tracers
Index Identify
ary data
case Contacts
Notify Contact/
contacts Follow up
testing
yes
Close Contacts
yes Prese no yes
Test Tested
nt for exit
+ve? -ve?
test?
no
20Summary
Contact tracing requires human contacts to be identified
Bluetooth apps use phones as a proxy for humans
but may miss transfers via surfaces
QR code apps tie individuals to locations, not to other humans
further steps required to link efficacy depends on these apps recording “in and out” can but human skill still needed to
humans at the locations location-based apps isolate individuals of interest assess risk as all premises differ
21You can also read
