The CoEpi app vision
To fully unleash the power of community epidemiology (CoEpi), we are creating an app that empowers people to leverage their mobile phone's technology for better awareness of granular infection risks.
The CoEpi app will only ask individuals to contribute their symptom data for the good of their community if they become unwell (experience symptoms); even when this occurs, the system is designed to permit only the minimum necessary information to be revealed, such that the actual identity of users is maximally concealed.
The CoEpi mobile app will ask permission for everything it does, and will allow individuals to:
simply install the app to get started, without disclosing any personal information.
enable or disable on-demand or background scanning for other compatible app users nearby using Bluetooth Low Energy (BLE), and recording (only on the local device) the history of other devices seen by the app.
be alerted, via push notification or by periodic anonymous polling, when another CoEpi user who has shared symptoms was nearby at any point when they may have been contagious.
When tracking symptoms, the app will optionally prompt the user for additional updates on symptom progression, providing the user with a comprehensive illness history if the symptoms progress to the point of requiring medical attention.
invite anyone to begin using the app (and optionally, register as a close contact) via text message, email, QR code, custom shareable URL, etc. Users would be encouraged to invite people in their own social network, as the app will be most useful when the people you’re most likely to be sharing germs with are using it.
In the future, our vision for CoEpi involves developing additional features, such as:
enable or disable tracking their own location history (only locally on their own device), with the ability to enable or disable such tracking on a scheduled, geofenced, or manual basis to avoid collecting any sensitive information the user would never wish to share.
produce a report on the device which shows a timeline of personal location history, likely as either a timeline or a map. This report should be able to be emailed or printed on user request. The use case of this information is in its application to individual health care, and in manual contact tracing activity should an individual wish to still contribute to community good after diagnosis but decline to use the networked methodology below.
be alerted when other CoEpi users come into close range for an extended period of time, and if both users consent, allow them to register a close-contact connection between the devices, allowing either user to preferentially share more information with trusted close contacts than with acquaintances.
consent to share any symptoms/results with close contacts, weaker recent contacts, future contacts, and/or public health officials.
In the future, we may also work with public health officials and enable an opt-in voluntary sharing of symptoms for population-based public health analysis, to visualize the distribution of suspected and actual cases of all respiratory and GI illnesses being actively transmitted in the community, most notably COVID-19. Global application of this or a similar and compatible system would dramatically improve the data that is made available to decision makers. If successful this system could be re-used for more effective control of existing transmissible infectious diseases such as influenza and measles, as well as future pandemic diseases.
Symptom reporting and sharing
Once fully integrated with secure and anonymous cloud APIs, the full power of CoEpi will come from the ability to report and share symptoms, and perform automatic real-time contact matching and tracing at scale. Here’s how it would work from a user’s perspective:
When a user enters any symptoms, they will be asked to report the recollected date/time of symptom onset, the current severity of symptoms, and any additional details based on the previous answers. This will generate exposure alerts to devices that they may have interacted with in close proximity during the time period when they were likely to be infectious, including a few days prior to symptom onset.
CoEpi's initial app versions focus on generating exposure alerts based on observed device interactions via Bluetooth. However, in the future we also envision the ability for CoEpi users to proactively opt in to alert their close family and friends and/or opt in to sharing with public health officials. Public health officials (and potentially epidemiological researchers) could choose to receive anonymized symptom reports from CoEpi users who optionally choose to share them.
Exposure matching and alerting
When someone who was in proximity of a potentially infectious CoEpi user is notified of the fact, they’ll be told about the details of the symptoms reported so far and the date/time of their proximity with the potentially affected user. For more severe symptoms consistent with COVID or influenza, they may also receive recommendations to begin practicing social distancing measures, working from home if possible, etc., and instructions to call their healthcare providers for guidance on what to do next. In future versions of CoEpi, the user will then be able to optionally enable periodic check-ins where the app will ask the user to report any symptoms they’ve noticed, what if anything they’re doing to reduce possible re-transmission, and will prompt the user to consent to share that information with other CoEpi users and/or public health officials if appropriate.
Completely free and open source
The CoEpi app and server code will be completely open-source and free to use.
There will be a main publicly-available server run by the CoEpi community. This server will endeavor not to retain any PII from individuals. Client IP addresses will not be stored in a database or recorded in log files any longer than necessary. There will be no “user accounts” on the server with usernames and passwords, profiles, or other account information: if someone gets a new phone, they’ll download the app again.
Additional details on our current work-in-progress implementations, data models for future functionality, and more can be found in our GitHub repositories.