Defence Force report applauds CovidCard despite Nelson trial bugs

Friday, 2 October 2020

Creator of the CovidCard in NZ talks about the wireless technology

The Bluetooth CovidCard has impressed the Defence Technology Agency which said in a report that it could potentially benefit contact tracing.

An earlier trial of the Covid tracing device overseen by Otago University at Nelson Hospital in May was hampered by bugs but indicated the device could accurately detect “close contacts”, another Government-funded report said.

The Government has said it could roll out CovidCards to improve contact tracing, but Government Digital Services Minister Kris Faafoi has said making them compulsory would be “an extreme last resort”.

CovidCard supporter, Trade Me founder Sam Morgan, has stepped back his involvement in the initiative after becoming frustrated with the Ministry of Health which he did not believe was taking it seriously.

* Singapore move 'a validation' of NZ's CovidCard approach, says developer

* Sam Morgan gives up on CovidCard in frustration with Ministry of Health

The Government has finally released its first reports on the CovidCard which are broadly positive but not conclusive about the technology.

* CovidCards work but making them mandatory would be 'an extreme last resort', says minister

The Defence Technology Agency (DTA), a branch of the Defence Force, said in its report that the CovidCard could benefit contact tracing by reducing the time taken to identify close contacts “if there is sufficient user acceptance and the system proves reliable in practice”.

“We believe that CovidCard has the potential to detect and record close contacts to a level useful for contact tracing, but that an unknown and possibly significant number of false positive contacts will be generated,” it said.

Early bugs in CovidCard’s firmware in May impacted the amount of research Otago University was able to conduct in Nelson.

The DTA said it was impressed with the quality of the effort put into the development of the CovidCard “and in particular the attention paid to security and privacy considerations”.

Based on tests of 12 prototype cards and documentation it believed the devices would be able to work for at least a year on their built-in battery, it said.

“The security architecture of the card has been carefully designed and we believe it meets reasonable privacy expectations.”

CovidCards are designed to be worn as a lanyard and automatically store details of other cards they come into close proximity with, creating records that can be accessed by manual contact tracers to identify close contacts if someone is subsequently diagnosed with Covid-19.

The DTA said the cards should be able to be carried in a shirt pocket, instead of worn as a lanyard, while still providing “a useful level of performance”, but that should be tested.

The devices were trialled at Nelson Hospital early in the Covid crisis in May when New Zealand was still in its first level 3 alert.

A separate report on those trials produced for Internal Affairs by Otago University’s Department of Public Health indicated the cards were substantially better at recording close contacts than manual records kept by the 42 people involved in the trial.

“The Bluetooth card registered a median of 18 close contacts per day compared to only three per day for the self-reported surveys.”

But the academics indicated that the lack of a good yardstick from self-reported survey data and a limited number of alternative measuring devices against which to measure the performance of the CovidCard prevented them from reaching any “substantial conclusions” about their effectiveness.

The trials were also impacted by issues with the card’s firmware which limited the amount of data available to the researchers, they said.

Both the Otago University and DTA reports were released by the Department of Internal Affairs on Friday after being requested by Stuff under the Official Information Act.

Engineer Dean Armstrong, who developed the CovidCard, said the cards used in the Nelson trial had used third-party hardware and firmware that had been supplied “really early in the piece and stood up in a heck of a hurry”.

“If we had developed the firmware there would have been bugs as well at that stage,” he said.

“One of the things that became clear was if we were going to provide the security and quality assurance that would be needed for an all-of-population roll-out then we actually needed to build this system.”

Armstrong said he was very satisfied with the DTA report, which was produced in July, and its feedback.

“They identified some good things, but overall it is pretty positive view of the work we have all done,” he said.

The Ministry of Health is believed to hoping to run a further trial in Rotorua, in part to test how people would feel about wearing the devices.

Stuff has requested the terms of reference for that study, but it has not so far been supplied by the ministry.