App for COVID-19 contact tracing

by Nicholas_Joseph 6 min read22nd Mar 20209 comments

5


Overview

This is a proposal for an app that would allow users to log/track their interactions, enabling faster and more effective contact tracing in the event that they contract COVID-19. Current methods for contact tracing are manual and rely on human memory, so this tool could greatly improve our ability to track the spread of the disease and communicate vital information to people at risk. In particular, I think that building this tool could be an effective use of time for EAs, especially those with web/mobile development experience.

There are a few things I’d love to receive help on:

  • Does this seem worth doing? In particular, are there difficulties or drawbacks that I may have missed?
  • There is an open questions section. If anybody knows the answers to those, I would appreciate it.
  • If you are a web/mobile developer and want to contribute, or know anyone who is, please let me know.

Motivation

This can help reduce the spread of COVID-19 immediately.

The current plan that I believe most governments are implementing is to impose strong shutdown and shelter-in type measures to suppress the disease until case numbers diminish. At that point, they will attempt to reopen things while maintaining low case numbers until a vaccine is developed, likely in ~18 months. Contact tracing is one of the main tools in use to keep the number of cases low.

Additionally, contact tracing is primarily an information and communication problem, which seems like something that technology can help with a lot. Therefore I think it’s a very promising and tractable opportunity for those with programming experience to contribute, and the EA community has many people with that background. Most other ways to help with COVID-19 require medical training or experience in manufacturing medical devices such as ventilators.

This could be helpful with future pandemics.

Unfortunately, COVID-19 will likely not be the last pandemic humanity faces, and despite the potential to have a horrific death toll, it is still at a lower scale than some of the catastrophes EA has been concerned with. My hope is that this will help in the long term by:

  1. Improving infrastructure for contact tracing, which seems likely to be useful for many infectious diseases, natural or synthetic. If this had existed beforehand, I think we would be in a better place.
  2. Enabling us to collect data on disease spread and use that to understand it better, assuming we can do this in a way that respects privacy.

What is contact tracing?

Contact tracing is a common public health tool, typically consisting of three steps, starting when a patient is diagnosed with a communicable disease:

  1. Interview the patient to identify all people they recently interacted with (going back 2-3 weeks in the case of COVID-19)
  2. Call all of these contacts and ask them to isolate themselves.
  3. Monitor all of them in case they were infected with the disease.

For more information, see the WHO site or the Wikipedia page.

Minimum Viable Product

The proposal is to make an app that would record contacts in real time to enable easier, faster, and higher recall contact tracing once somebody is sick. Since I think it is quite urgent to get this out quickly, I will describe what I think is the minimum viable product (MVP) below and then describe additional features after.

The MVP would be a web application that does the following:

  1. Prompt a user each day to manually record everyone they have interacted with that day (likely from a list of FaceBook friends or phone contacts). They can also list people they interact with every day to avoid constantly updating.
  2. If a user is diagnosed with COVID-19, they are able to notify everyone they have interacted with for the past three weeks with the click of a button. This notification can be anonymous and should come up with links to instructions on how to safely deal with this situation.

This would help by making contact tracing:

  • Easier: They would only need to click a single button instead of manually calling lots of people. Since public health officials may be too overwhelmed to do this properly, this could increase the amount of contact tracing happening significantly.
  • Higher recall: I would find it hard to remember everyone I have had contact with in the past three weeks, particularly when sick enough to be in the hospital. Doing this ahead of time should make it more complete.
  • Faster: No time is wasted waiting on possibly overwhelmed officials to do the contact tracing, it can be sent out as soon as a diagnosis is received.

Additional Features

I think the above will provide some help and require relatively little from a user, but there is also the opportunity to improve this with more features.

Location tracking

The most obvious is to make this into a phone app and use location tracking to determine everywhere that people have been. This is currently being used successfully in countries like China, South Korea, and Singapore. A less intensive form of this would just ask the user to manually add specific stores or crowded locations they have been in and at what time.

The above location could then be used to make a map of what areas are dangerous and have had people with COVID-19 near them, likely expiring after three days. South Korea has a map similar to this.

Assessing risk level beyond first-order contacts

Traditional contact tracing only considers first-order contacts, probably for reasons of efficiency. However, this app could inform people if they are connected to someone with COVID-19 in second degree connections. Other risk factors could also be considered to give an overall risk rating for a user.

Report symptoms and send less confident warnings based on that

There is also a shortage of tests in many locations, so many cases are likely going undiagnosed at first. The app could also query for symptoms and send out less confident warnings if recent contacts have fever/dry cough or other common indicators of COVID-19.

Gamification types of things to increase adherence to recording

In order to make sure that people really do enter their contacts each day, we could make it visible to others how long it has been since they updated, send texts if someone lists them as a contact that they didn’t list, etc. I’m sure there are useful lessons to be learned from exercise/diet apps that require the user to enter all their workouts or meals as well.

Concerns

There are certainly ways this could go poorly or not be that helpful, and I would love to hear from anyone who has ideas or concerns I have not considered.

Privacy concerns

Users may have a very strong aversion to giving their location at all times to an unofficial and untrusted app. This is understandable, but I imagine there are a number of people who are willing to sacrifice some of their privacy in an emergency to protect their health and that of others. It would definitely be important to be very transparent on what data is being collected and make sure the user has opted in to sharing that.

Others are already working on this

There are many cases of this being worked on already. South Korea, China, and Singapore all have similar cases, mostly run officially by the government. The WHO is currently working on an official app that will do some of this.

I think there are a lot of advantages to having an official branded app, and could imagine discontinuing or merging this project if one of those is successful. However, at the moment, I believe that most people in Western countries are *not* tracking their contacts.

Of the apps I am aware of, the WHO app is the only one that seems targeted for the US or Western countries. It’s v0.1 does not include contact tracing and is primarily focused on providing information. However, it is intended to work across lots of cultures and might need to make strong compromises with privacy in order to have broad adoption and protect its brand (e.g. it would be very bad if people lost trust in the WHO if its app was poorly received). It could be beneficial to have multiple apps with separate brands so that people with different privacy preferences can use different apps.

There are some advantages to developing multiple apps in parallel. First, many tech projects can fail, be delayed, or switch focus to new priorities. Given the consequences of COVID-19, having some redundancy may be a worthwhile precaution. It could also be good if multiple apps try different approaches and learn lessons about what works well and what doesn’t. It is always possible to stop working on this app and redirect users to the official app if it turns out to be much better.

Sign up and follow-through

Perhaps not enough people would sign up for this, or once they do sign up they would not consistently report their contacts. One thing to note is that even if only a small number of people sign up, this can still be helpful. Users can send out texts to contacts even if those contacts don’t have the app. But of course, the more people that use it, the more of an effect it would have.

Would this cause undue panic?

A bad actor might list all their contacts and then falsely mark themselves as having COVID-19. This could then cause many people to panic and isolate themselves or to weaken trust in the warnings they receive. Alternatively, even without dishonest reporting, people may overreact to the notifications and do more than what is recommended. Some of this can be controlled by limiting the number of contacts and careful monitoring, but is still a concern.

Can we get this out there?

The Apple and Android stores are currently blocking unofficial COVID-19 apps in order to curb misinformation about the disease. I proposed starting with a web app for this reason, but it could be very difficult to publicize it. I think if a web app is showing promising results, it might be possible to partner with an official organization to get an app published, and if not, the web app might be sufficient (e.g. running in a browser on a phone).

Open Questions

Is this legal with regards to privacy/medical information sharing?

If anyone here is a lawyer and has expertise on HIPAA or other relevant regulations, I would love to get advice on this. I’m particularly interested in potential roadblocks and solutions or alternatives if you have ideas. I have no legal experience and don’t know how much of a blocker sharing location or medically-related information will be even if people consent to it.

How important is the location mapping aspect of this?

I think the main consideration here is what proportion of transmission is to random passersby (e.g. coughing near someone on the street or touching the same surfaces) vs. someone you know and would be able to record. I am interested in seeing more data on this.

How hard would this be to create?

I’m a software engineer, but I mostly do machine learning and haven’t done much web or phone app work. It seems to me like a reasonably simple app to build, but I might be unaware of some technical blockers.

Related Information

Disclaimer

I apologize if this post has some errors. I did my best to check things, but prioritized getting it out fast for quick feedback given the urgency of the situation and how fast things are changing.

5