The pitch from technology companies goes something like this: We can tap phone data to track Covid-19 infections in U.S. communities and swiftly warn people about potential exposure, all without ever compromising anyone’s privacy.
Apple and Google turned heads a few days ago when they announced a joint effort to bolster this public health service — a task known as contact tracing — by building software into smartphones that relies on Bluetooth technology to track users’ proximity to one another. Facebook is participating in a similar effort led by the Massachusetts Institute of Technology.
While their brands give them instant credibility in the business world, infectious disease experts aren’t convinced the technology offers a tidy solution to such a complicated public health problem.
“Some of the promises of what are sometimes called proximity tracing [tools] may be a little bit over the top,” Tom Frieden, a former director of the CDC, told STAT when asked about such efforts. He added that it’s not yet clear whether digital surveillance methods used in several Asian countries truly helped to control the spread of infections.
“What we’ve seen from various countries is information that this was done, but not information that it was effective,” he said, adding that any proposal to tap phone data in the U.S. must include a “completely transparent sharing of what’s being done [and] why it’s being done.”
Other countries are using electronic bracelets to track users’ movements or even automated systems that classify citizens’ health status to determine whether they should be isolated or allowed entry to public places.
The great promise of technology during the pandemic is that, in many domains, it can make the delivery of health care safer and more efficient, and can guide the work of public health experts scrambling to track the contacts of those infected. But the value of these digital tools has not been established.
Here are five burning questions to keep in mind as these efforts move forward:
Will smartphone-based tools really improve contact tracing?
It may be technically possible to use phone data to help identify possible exposures, but that doesn’t mean it will improve the identification of cases and result in better containment.
That’s because contact tracing is not a purely technical exercise. It requires public health workers to navigate sensitive interviews with infected and exposed patients to extract valuable information. In theory, using cellphone data could help jog peoples’ memories to identify exposure incidents, but this potential benefit has not been established in practice.
The use of technology could interfere with a public health worker’s ability to build trust with infected patients, and in turn, undermine their ability to deliver supportive services and guard against further transmission.
“Some of the best practices [of contact tracing] include ensuring that people have food in their home, that they have security, that they have the ability to care for elders or children who may need to be removed from the house,” said Frieden. “Public health works by making the patient the VIP of the program, and that’s what we need to do here.”
Frieden also said public health workers are trained to recognize circumstances where contact tracing requires a deeper inquiry to identify secondary contacts, which is especially important when dealing with a virus that can be spread by asymptomatic individuals. In those cases, a digital tool wouldn’t cut it.
“You have situations where there is a complexity and you need to extend the investigation because it was a choir or a nursing home or some other congregate facility,” he said. “And there you really need the highly specialized disease detective epidemiologist working with you.”
How do you guard against bias and discrimination?
Not everyone has access to a smartphone or can keep up with their monthly bills, especially at a time of skyrocketing unemployment. It’s possible that efforts to leverage location data could work much better in wealthier communities than in places with higher levels of poverty and less smartphone penetration.
That could not only perpetuate bias, but undermine the effectiveness of the intervention in the first place.
And even if information displayed by an app is anonymized, it could still be used in ways that the developer did not intend. If an app showed that infections were concentrated in an ethnic neighborhood or that exposure happened at a particular religious institution, people belonging to specific groups could be subjected to discrimination or even violence.
“These things have to be thought through very carefully.”
Madhave Marathe, University of Virginia researcher
“We’ve seen that effect happening already,” said Madhav Marathe, a computer science professor at the University of Virginia who studies the use of advanced computing techniques in epidemiology. “These things have to be thought through very carefully.”
How accurate are the tools?
Geolocation data from phones is generally accurate, but is it accurate enough to confirm that you’ve come in close contact with an infected person? One person may be separated from another by a wall in an apartment building or an aisle in a grocery store, but their phones would show them in close proximity. Being in “close proximity” doesn’t always mean a person should be tested for Covid-19 or automatically self-quarantine.
Phone software also does not take into account relevant environmental factors, such as the amount of ventilation in a room where contact may have occurred.
Jay Stanley, a senior policy analyst with the American Civil Liberties Union, said phone data could be helpful by giving public health agencies a general sense of the level of infection in their communities, and help hone their advice to the public. But he said that such apps would need to be carefully crafted to ensure they’re not overly prescriptive.
“It’s not sufficiently [accurate] to do this in an automated way where you take a giant dataset of millions of GPS coordinates and you cross reference them with the GPS trails of infected people,” he said. “That’s highly impractical.”
And, he added, no matter how accurate phone tracking software might be, it still comes with privacy concerns.
“Whether it’s GPS, Wi-Fi, or Bluetooth, it is plenty accurate to invade privacy and reveal your general location, but it is not accurate enough to determine to whom you may have been exposed,” Stanley said.
Will enough people get tested — and voluntarily download a tracing app?
Effective contact tracing requires adequate testing to determine who’s infected. But communities nationwide do not have enough supplies or staff to test everyone, leaving them unable to track the number of infections.
In hard-hit areas such as New Jersey, only about one of every 75 residents is getting tested, leading to long backlogs. Public health experts hope that testing will become more widely available before a possible second wave of infections in the fall, but there is no guarantee that will happen.
The accuracy of digital contact tracing also relies on peoples’ willingness to download an app to share their information with public health authorities. Unless apps gain adoption from a large share of citizens — both healthy and unhealthy people — it will be difficult to accurately track the spread of the illness.
In Singapore, officials launched an app called TraceTogether that also relies on Bluetooth technology to notify users of contact with an infected patient. As of March 23, about 620,000 people had downloaded the app, according to the Facebook post of a top government official. That seems like a high number, but some onlookers have suggested that this level of uptake would only account for a tiny percentage of random contacts with infected individuals.
In the United States, neither tech companies, nor government officials, has suggested the possibility of mandating the use of an app, but that might be the only way to ensure widespread adoption.
How do you balance privacy and public health in a pandemic?
This is the elephant in the room. Digital contact tracing involves a level of surveillance that could make a lot of people uncomfortable, especially given the involvement of large technology companies with spotty records on privacy.
Apple, Google and Facebook are asserting that Americans do not have to choose between health and privacy during the pandemic. In their contact tracing efforts, they are proposing to scramble the data so that information can be gleaned from the app without revealing the identities of the participants. In announcing their effort, Apple and Google also released technical documentation explaining how they will encrypt the information to guard against malicious attacks.
While anonymizing the data is technically feasible, watchdogs in Congress have raised concerns that location data could reveal other sensitive aspects of users’ lives, including religious affiliations or political preferences. Sen. Edward Markey (D-Mass.) sent a letter to the White House Office of Science and Technology Policy asking the Trump administration about how data gathered through partnerships with tech companies would be collected and used, who would have access, and whether it could be later de-anonymized.
“We need assurances that collection and processing of these types of information, even if aggregated and anonymized, do not pose safety and privacy risks to individuals,” he wrote.
Apple and Google have said that they will not collect location data or other personally identifiable information from users. But it remains to be seen whether the federal government and its potential technology partners will be fully transparent about the use of the technology, and explain it in ways that citizens can easily understand.