A secure and privacy-protecting data platform to track the spread of COVID-19
Developed in collaboration with Rutgers University, sponsored by the National Science Foundation, National Institute of Health and Rutgers IDSS.
Visit Website
Company Background
Existing COVID tracking techniques by governments ranged from authoritarian with mandatory quarantines, lockdowns and forced tracking and reporting of symptoms, as in Australia, China, and the UK to self-reported symptoms and self-enforced social distancing as in Pakistan.
Various overt and covert techniques were employed to profile whether an individual may have been exposed to COVID-19 exposure, including tracking user locations and search histories, etc. Due to potential ramifications, people tended to not self-report symptoms where they could. As a result, this greatly reduced the ability of local governments and medical facilities to prepare and provide a coordinated response.
The Institute of Data, Society and Systems at Rutgers University was awarded a grant by the National Institute of Health and the National Science Foundation to study how to reduce the reluctance of people to self-report symptoms through anonymity guarantees.
The challenge
This time-sensitive project required quick turnaround and deployment as the pandemic quickly became a health crisis with millions of lives at stake. The client wanted to build a secure crowd-sensing platform that allowed users to share anonymized location and symptomatic information while also maintaining their privacy, and have a real-time dashboard that would help people track the spread of COVID-19 as it progressed. For maximum accessibility, the platform needed to be available as a web app and also a mobile application on the Apple and Google app stores.
The goals of this project were three-fold:
- To build an application that tracks the spread of COVID while preserving the privacy of people reporting their daily well-being and potential COVID symptoms, making sure that public concerns about sharing of sensitive and potentially identifiable data are minimized.
- To build a database of voluntarily-reported COVID symptom reports for further research into the feasibility of self-reporting symptoms and the impact of privacy guarantees on the willingness of people to share personal health information.
- To build a modular web and mobile app that can be used in the future for privacy-preserving voluntary public data collection, especially with regards to disease outbreak tracking.
The time constraint for rapid development and deployment to production for the app was a fourth objective due to the pressing circumstances caused by the rapid spread of COVID-19 across North America.
They were able to turn around an entire platform within 6 weeks. Kudos to their responsiveness and speed!
Hafiz Asif
Assistant Professor, Hofstra U. / Postdoc Researcher, Rutgers
How we approached this project
The onboarding was designed so that no personally-identifiable information was collected. Demographic information was input as ranges which were deliberately structured to request just the required degree of precision to provide meaningful data for analysis, without enabling profiling.
To provide insight into how much information people are willing to share, as part of a user’s onboarding they are asked to select their preferred level of anonymization using a slider. This slider impacts how vaguely or precisely their location, demographic and symptomatic data would be recorded.
The user is then shared an initial questionnaire about how they’re feeling and if they have any potentially COVID-induced symptoms. This questionnaire is then sent to the user every day using browser notifications on web or push notifications on mobile, and users are given the option to skip it. If a user skips the survey at one time, they are prompted to fill it out for that day one more time in the evening.
The user dashboard allows the user to toggle between two map-based views which are centered on the user’s location by default. One view shows COVID-19 case data from the COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University, at the national, state or county level depending on the degree of zoom of the map.
The other view shows a ‘viewfinder’ experience which shows the user how many people reported themselves as having no or some COVID-19 symptoms within the visible area over the last 14 days. We used differential privacy to protect users’ location data and anonymize data points. Differential privacy is a mathematically sound approach to anonymizing results of queries from a dataset to ensure that the individual data points are not reflected in queries.
- The dashboard is an interactive map which users can navigate through easily to view the data of any area in the US they want. This makes data easily accessible and easy to understand.
- The daily report questionnaire has a simple, practical design and MCQ questions to make it efficient and time-saving for users, and to make data collection and analysis faster and easier.
- Differential privacy integration allows users to pick their preferred privacy level, while also allowing their data to be useful for analysis.
Flutter was selected as a robust cross-platform framework for the mobile application, whereas the web application was built in the ever-popular React. Both were selected for their high levels of support and ease-of-use. The web server was developed in NodeJS.
As the data models and database schema was iterated repeatedly throughout development, MongoDB was selected as a highly-flexible database. The data server that managed data processing, updates, anonymization and backups was developed in Python, leveraging NumPy and Pandas for various functions.
A separate WordPress application is used to manage the platform’s static pages for easy content editing. All of the servers and the web application were built on the AWS EC2 infrastructure.
The result of our work
We engineered COVID Nearby as a platform where people can anonymously and securely report their location and health symptoms, and generate information that is crucial to fight COVID-19. This information is used to identify the presence of the virus and track its spread in US communities.
COVID Nearby was able to track and gather data from over 34,000 users for further analysis over a period of six months.
This is what we achieved for Covid Nearby
The project received additional funding by the National Institute of Health to modularize it to build a privacy-preserving crowd-sensing platform that could be instrumental in future public data collection efforts at the university level and potentially across academia.
Further study continues into how privacy guarantees affect the willingness of people to share information that may help in coordinating service delivery and emergency responses by local governments.
01
A dynamic website that compliments their branding and service
02
Created a user journey that makes finding hot zones simple and easy to navigate around
03
Implemented a notifications and alerts module to improve end-user safety and precautions
04
Deployed mobile apps on both stores to ensure that information on nearby outbreaks is always available
34,000
Users self-reported symptoms
2022 Best Paper Award
IEEE Computer Society
3
Research Publications
Let’s make something new.
Tell us about your product, idea or what’s on your mind. If you’re not yet sure, a simple “hi” will do. We know what questions to ask to figure out how we can collaborate.
