Resolving the $100 million TraceTogether dilemma

TraceTogether
Image: GovTech

Japan recently announced that it will soon launch its own contact tracing smartphone app. According to the Straits Times, to protect users’ privacy, the app does not collect names, phone numbers, user locations or any other personal information. This follows the launch of SwissCovid, an app built in Switzerland which boasts of similarly strong privacy protections.

Both apps use software jointly created by Apple and Google, called the Exposure Notifications System, which uses Bluetooth technology to help health authorities perform contact tracing, while ensuring that user privacy and data security remain central to the design.

Singapore’s Government Technology Agency (GovTech) built its own TraceTogether app on a different protocol. Jason Bay, TraceTogether’s product lead at GovTech, wrote in a blog that since March, GovTech had been working with Apple and Google on the specifications for contact tracing technology which “allows cross-border interoperability.” He wrote this on April 10th, the day that Apple and Google announced their partnership to create the Exposure Notifications API.

It is widely known that TraceTogether does not work well on Apple’s iPhones, which are used by a third of smartphone users in Singapore. This is because iOS, the operating system running Apple devices, suspends Bluetooth scanning when the app is running in the background, preventing it from collecting data from contacts in the proximity of the user.

Mr Bay posted as recently as April 20th that GovTech will continue to work with Apple and Google to improve TraceTogether using the new API. It therefore came as a surprise when on June 5th, Dr Vivian Balakrishnan, Minister-in-charge of the Smart Nation Initiative, told Parliament, “We have had repeated discussions at both the technical and policy level with Apple, but we have not yet been able to find a satisfactory solution.”

The government has not announced any further plans to use the Apple-Google API. Instead, it plans to issue a wearable device to all Singapore residents and make it mandatory for everyone to carry it around. On June 13th it was reported that the government had already awarded a tender to a company to manufacture the first 300,000 units of the device. Based on the tender price for the pilot batch, CNA estimated that the device will cost $110 million to be rolled out nationwide.

Why couldn’t TraceTogether be made to work with the Apple-Google software? Dr Balakrishnan did not elaborate on what the technical and policy issues were. However, an examination of the two main protocols used by contact tracing apps around the world could give us some clues.

Centralised vs decentralised contact tracing

TraceTogether uses BlueTrace, a centralised report processing protocol. When a person tests positive for Covid-19, they will be required to upload the entire contact log from their phones onto a central server managed by the government for the purpose of contact matching and tracing.

TraceTogether’s current protocol also has no special privileges over normal apps, preventing the app from running Bluetooth scanning in the background.

Contact Tracing
Image: BBC

Exposure Notifications, on the other hand, employs a decentralised report processing protocol. User data is not stored in a central server and the matching of Covid-positive cases is done on users’ devices. Users can still opt to share their phone number and details of their symptoms with health authorities through the app, so that they can be contacted and receive advice on the next course of action to take.

Because the Exposure Notifications protocol is implemented at the operating system level, it allows for more efficient operation as a background process. It could also extend battery life and improve detection across iPhone and Android devices.

Several countries which initially pursued the development of centralised contact tracing apps have now decided to adopt the decentralised architecture pushed by Apple and Google. In late April, Germany abandoned a home-grown app design in favour of a “strongly decentralised” approach after Apple refused to budge on the settings of its iPhones.

At least 22 countries have now received access to use Exposure Notifications, including Austria, Belgium, Germany, Ireland, Italy, Latvia, Poland, and several U.S. states.

Even Australia, which created its COVIDSafe app based on TraceTogether’s open source code, is now exploring a switch to Exposure Notifications.

Why do we need a contact tracing app?

With all the privacy risks that contact tracing technology brings, why do we even need to use it? Don’t the traditional modes of contact tracing suffice?

Effective contact tracing and widespread community testing have been recognised as the twin keys to enable the safe and wide opening of economies from Covid-induced lockdowns, of the sort we have endured in Singapore for the past three months.

Manual contact tracing involves interviewing the patient and asking them to recall all their movements for the past 14 days to determine whom they came into close contact with. Large teams of contact tracers then track down these contacts to instruct them to self-isolate or get tested. This is a laborious exercise which can take a few days. The delay could result in an infected person roaming the streets for several days, unknowingly transmitting the virus to others.

Contact tracing apps, if widely adopted, can instantly alert all the close contacts of a new case and instruct them what to do next, drastically reducing the time needed for contact tracing and stopping the spread of infections.

Contact Tracing 1
Contact Tracing 2

The win-win solution for Singapore

The Covid-19 pandemic has done much damage to our economy, with as much as a 7% GDP contraction expected in 2020. It is the worst recession in our nation’s history. Therefore any solution that will allow us to safely reopen our economy should be considered. Contact tracing apps can provide a way forward.

We need not make a false choice between privacy and public health. The Exposure Notifications System provides both privacy protections and privileged operating system access to allow contact tracing apps to work on almost all smartphones, even when they are locked.

By improving privacy protections and reducing the battery drain on phones, an enhanced TraceTogether app will become more attractive for Singapore residents to install on their phones, improving its current 25% adoption rate to a level closer to what is needed for effective contact tracing.

As more coronavirus-torn countries manage to contain their outbreaks, talks are underway on the establishment of “travel bubbles”. This would allow quarantine-free movement across borders. An important prerequisite would be effective contact tracing.

Interoperability between contact tracing apps would certainly help. As more countries around the world decide to adopt the decentralised protocol created by Apple and Google, it makes sense for Singapore to move in sync with them, to enable TraceTogether to “talk” to apps in other countries.

Given that 91% of the population in Singapore uses smartphones, rolling out a working contact tracing phone app can be done much more easily and cheaply than issuing a hardware dongle to all 5.7 million residents. It is also more seamless to fix bugs or add enhancements to an app and roll out an update to App Store or Play Store. If the hardware token has a serious bug, that’s $110 million down the drain.

It is still not too late to reverse course. The TraceTogether app should be re-programmed to adopt Apple and Google’s Exposure Notifications system. This will enable the app to run effectively on all smartphones, maximise adoption, protect privacy, enable cross-border interoperability and, most importantly, become a real weapon in our battle against Covid-19. The wearable device then only needs to be issued to the 9% of residents who don’t own a smartphone, saving taxpayers over $100 million.

MOH’s “surveillance programme” for Covid-19 testing an encouraging development

Photo by Summer Chan on Unsplash

The headline “MOH flags concern over unlinked COVID-19 cases” may sound rather ominous, but MOH’s announcement of its “ongoing surveillance programme, where a small sample of patients at our primary care facilities are tested for COVID-19 infections” is an encouraging development. MOH added that they have “picked up some cases through these tests, which is an indication of undetected cases in the community.”

MOH did not elaborate on how this “small sample” of patients is selected. However, it is likely that not all of them present the typical Covid-19 symptoms and risk factors, but they were still tested as part of this programme. As I had explained in my posts last week and and this week (links below), it is important that the pool of patients tested for Covid-19 is widened beyond the symptomatic. This is because up to 70% of those infected with the disease may be asymptomatic. They could end up being “silent spreaders”, which not only infect others but also make contact tracing extremely difficult.

The only way to identify more of these undetected cases in the community is to do more community testing. This allows the authorities to isolate more positive cases and staunch the spread.

Also read:
https://jby.bwy.mybluehost.me/2020/04/community-testing-for-covid-19/
https://jby.bwy.mybluehost.me/2020/04/identifying-the-silent-spreaders/

Identifying the “silent spreaders”

Photo by Martin Lopez from Pexels

If up to 70% of those infected with Covid-19 are asymptomatic (or pre-symptomatic), then we need a way to identify and isolate these “silent spreaders”. Wearing a mask will help but it will be delaying the inevitable — they will eventually spread it to someone, whether at home or in their workplace.

This is where more widespread testing is needed. We could start by swabbing anyone who reports even mild symptoms and all the workers in the affected dorms (where active case-finding is already in progress).

If there aren’t enough test kits or chemical reagents, then let’s make it a national priority to source for them or manufacture them locally. This is a national emergency and time is of the essence.

Community testing for Covid-19

With a record 140 local cases of Covid-19 recorded yesterday (Apr 8th) in Singapore, it is timely to examine whether there are further approaches to containing the Covid-19 pandemic that can complement the so-called “circuit breaker”, which is in place until May 4th. (Update: There were 284 new locally-transmitted infections on Apr 9th, double the previous day’s record.)

One approach I hope the Ministry of Health (MOH) is considering is community-based testing for the coronavirus. This approach has been advocated by infectious diseases experts and is even being implemented in some countries like Iceland. However to-date, there has not been much public discussion about it in Singapore.

Photo by Chokniti Khongchum from Pexels

Community-based testing involves the testing of a large percentage of a country’s population for Covid-19, even those who may be asymptomatic or have only mild symptoms. By preemptively identifying individuals who are carrying the virus, we can isolate and treat them, and prevent further community spread. If done correctly, this may enable governments to slowly lift some of the draconian lockdown measures which are causing much pain for businesses and workers.

Singapore’s current approach to coronavirus testing is to focus on individuals with symptoms. When asked why testing has not been used more widely, the Director of Medical Services (DMS) said on April 6th that MOH’s approach is to test in a very targeted fashion so as not to waste the tests. He said that Singapore does not see the need for widespread community-based testing now, although it is “something (MOH) might review and consider doing in the future.”

The DMS said that Singapore has the capacity to continue its targeted testing approach. However, it is unclear whether we have the capacity for community-based testing. In the UK, a shortage of chemical reagents has limited the number of tests they can run a day. Could this constraint be the reason for Singapore’s current testing approach? It would be good if MOH could share the constraints it faces and its plans to overcome them.

Another concern about community-based testing might be the cost. Currently, each antigen test costs around $140, while the antibody test costs around $25. This cost varies from country to country. The antigen test (reverse transcription polymerase chain reaction or RT-PCR) tells if a patient currently has the virus, while the antibody (serological) test indicates if a patient had the virus. Tests are typically run a few times on each patient to confirm a negative result.

If community testing were to be rolled out nation-wide, we are looking at the order of hundreds of millions of dollars. This cost could be mitigated as more competing diagnostic test kits enter the market, leading to greater price competition. Indeed, the HSA is already expediting the approval of Covid-19 diagnostic tests.

The current “circuit breaker” in Singapore will last until May 4th. It may well be extended if infection rates do not come down. The economic and social costs of an extended lockdown could far exceed the cost of mass community testing.

We could begin mass testing on the foreign worker population living in the three dormitories which have been declared as isolation areas. Their crowded and cramped living conditions are breeding grounds for the spread of the virus. Many of the local transmissions reported in the past week have been linked to these dorms.

The government should test all the workers in those dorms immediately, even those who have not shown any symptoms, and quickly isolate and treat those who test positive. This will prevent the further spread of the coronavirus in the dorms. It will also enable those who test negative to be moved to other housing facilities. Many of these workers need to leave their dorms to work in essential services.

Photo: Kevin Lim, The Straits Times

Carrying out community testing on a nation-wide scale requires an incredible amount of coordination and resources. It is not something that can be rolled out overnight. Mass testing would involve, among other things, the development of digital technologies to coordinate the roll-out of the tests and the purchase of enough personal protective equipment to conduct the tests. Without adequate preparation, hospitals and clinics will be overwhelmed by test requests.

The government needs to start planning for community testing earlier, rather than later when the need becomes too acute. If we move too late, a global shortage of test kits could be a problem, as other countries snap them up faster than they can be produced.

Social distancing on public transport

Photo: The Online Citizen

Crowding in public transport is a weak link in social distancing measures that are being rolled out island-wide in view of the coronavirus pandemic. People are expected to sit at least 1 metre apart in the office and in restaurants, but then they proceed to squeeze into packed trains on the journey home. One solution would be to increase the frequency of buses and trains. That will reduce crowding and increase social distance (better than simply not talking, which one of our ministers suggested), and it is something within the control of the government and the public transport operators.

S’pore’s first Covid-19 deaths

So sad to hear about Singapore’s first two Covid-19 deaths happening within hours of each other. (Update 29/3/2020: A third death has since occurred.) My deepest condolences to the families of the two patients, an elderly Singaporean lady and an Indonesian man. This is something that we had been bracing ourselves for but the news still hits you like a ton of bricks. Let’s all continue to protect ourselves, our loved ones and the people around us by doing our part.

Bullying and racism in schools

Reading the original Twitter posts by the girl’s sister brought tears to my eyes and made my blood boil. What was not mentioned in the ST article was how the little girl was allegedly called a “black dustbin” and that she resorted to having her recess in the toilet because it was the “best place for (her) to cry and eat at the same time”. While details of these alleged incidents are still unclear and I’m sure the school and MOE are investigating, I think this should spark some soul-searching among all Singaporeans, especially us parents (my children are around that age). What values are we inculcating in our children? Do we harbour racist attitudes and inadvertently pass them on to our children? Do we teach our children to stand up to bullies, even if they are not the victims?

Straits Times article: https://www.straitstimes.com/singapore/bullying-in-schools-here-wrong-and-cannot-be-tolerated-ong-ye-kung

Allied educators in schools

Glad Leon Perera asked this Parliamentary Question (PQ). It arose after my conversation with a teacher during house visits in Fengshan. This teacher voiced concern that the number of classroom teaching assistants was being reduced. I was surprised to hear this and decided to look deeper into the issue together with my colleagues from The Workers’ Party. It resulted in Leon‘s PQ. He didn’t just stop when he got the initial answer, but continued to press the Parl Sec for the detailed numbers.

This highlights the value of the ground feedback we receive from residents. If you meet our MPs and volunteers when they drop by your place during house visits, please raise up such issues of concern to us so we can bring them to Parliament for debate. We work as one #TeamWP to be your voice in Parliament!

(TLDR: There are 4 categories of AEDs. The headcount in one of the categories, AED (Teaching and Learning), is indeed being reduced while those in the other 3 categories is being increased.)

Video credit: CNA