Contact tracing requires a strong understanding of location and context of encounters. Although Bluetooth technology does not provide location or context of the encounters, we would like to share our excitement and also some thoughts about the forthcoming Google/Apple Bluetooth Proximity API (GAB).
Safe Paths is open source, open standards, privacy-first and works closely with public health entities. Our approach is to roll out Apps, SDKs, privacy-preserving network backbones, and interoperable protocols so that any developer can communicate with any public-health data. The APIs we support come from many partners and we plan to include the Google/Apple Bluetooth proximity solution.
At the same time, we need to understand the role of various technologies at play, their context to the end-user, as well as to the health officials and communities.
Thus, proximity detection is a piece of the contact tracing puzzle and contact tracing is an even smaller piece of the public health puzzle. With location and context, we can develop a more holistic solution.
There are several reasons to consider a holistic multimodal solution that includes location tracking in addition to Bluetooth proximity ID. Any one technology has limitations of either false positives, or false negatives. Ultimately, the best solutions will leverage the advantages of GPS, Bluetooth, and WiFi in order to create more accurate, useful, and private data.
Adoption rate vs effectiveness: Bluetooth requires many people to use the apps for it to be valuable. In Singapore the Bluetooth App penetration is 12% which means only 1.44% of encounters are recorded (0.12*0.12). As a reference, Casey Newton has a good piece on Why Bluetooth apps are bad at discovering new cases of COVID-19. While adoption can be increased through government and big tech buy-in, these numbers still hamper effectiveness.
GPS based App scales linearly: With a 12% adoption rate, 12% of infected hotspots will be identified and nearly everyone in town will hear about them, eg. local news channels. GPS is useful even if smartphone penetration in a region is not widespread.
False positives or False Negatives: Using bluetooth alone has the risk of generating either false positives, or false negatives. For instance, if you are living in the apartment below an infected person, sharing a ceiling/floor, or working from a neighboring office, Bluetooth can still flag you as at risk. False negatives could occur if your phone was turned off or not with you when you came into contact with an infected person. This ambiguity can only be handled by a good UX.
Contextual information is key: Users do not like to get an alert without understanding the context. They need to know where exactly the encounter took place in order to trust the system. This location information can be provided by GPS location logging. Access to location information can also help human contact tracers conduct interviews by helping the patients, for example, remember if they were wearing a mask or not, whether they shook hands or not, etc.
The prime need of the health officials is their ability to call the exposed person directly to assess their personal situation and provide guidance about testing/isolation/hospitalization. Health care officials also need dashboards of emerging hotspots to be able to map the spread. Without GPS, it is difficult to create tools for heatmaps and spread analysis. Bluetooth does have certain advantages, especially in dense urban environments, but the optimal contact tracing solution will ultimately be multimodal (Bluetooth, GPS, and Wi-Fi), leveraging the advantages of each technology. For example, contextual information from GPS location logs can help a simple rejection mechanism for the false positives created by Bluetooth. GPS also allows for the creation of crucial virus heatmaps for health professionals, without needing large scale adoption by the population. Safe Paths is developing privacy preserving self-reporting that will not lead to misinformation and abuse. And self-reporting (heatmap as well as social circles) is very important as going to testing sites still has a lot of friction, especially among young people. While a Bluetooth API certainly makes aspects of contact tracing easier, public health officials need more than simply contact tracing to address the epidemic. They need a larger ecosystem that can help them with quarantine management, health verification, patient interview, hotspot identifying, and more. All of these tools furthermore need to protect privacy and be adapted for local conditions.
Contact tracing is just a tiny part of the public health interventions we need to build: for encounter memories, checking on loved ones or checking on co-workers you meet daily etc., quarantine management, health verification, sick leave certification and so on especially for restarting the economy. The tracing solution apps should be able to amplify the role of Public health officials.
SafePaths is, above all, about building open standard end-to-end solutions for citizens and public health while maintaining privacy and scalability. We will continue to use the best tools, Bluetooth APIs included, to achieve this aim. In the early phases, the emphasis is on rapid iteration and deployment for solutions for epidemic tracking. In the later phases, the goal is building encrypted computational methods that can be useful in any future societal disruptions. Available here is Prof. Raskar’s 2019 talk about creating an honest impartial broker to address challenges in a fragmented society.
We are already developing pilots in 30 jurisdictions across the world, with Ernst and Young (EY) as a deployment partner. SafePaths is working closely with public health entities, and we plan to roll out apps and end-to-end solutions for each entity that makes use of various technologies including the recent Apple/Google Bluetooth API. We are furthermore building interoperable protocols and standards, so that each of these jurisdictions can benefit and learn from each other.
We want to stress that Apple and Google are releasing Proximity AP but not a full contact tracing solution. When the APIs are released, which is slated for the coming months, it will give contact tracing tools such as Safe Paths the ability to work at a deeper level on phones, improving battery life, effectiveness and privacy.
Here, it might be important to bring to light that while there are many contact tracing solutions that could be built on these APIs, there are more privacy issues with Bluetooth than GPS based solutions because each phone (not just the infected person’s phone) is emitting the Bluetooth ID every few seconds. So a company that has hundreds of millions of users can easily change their app to listen in on broadcast IDs that change very slowly (15 mins). Smartphones can help reduce the spread of the virus but any network analysis could cause unacceptable intrusion to privacy and human rights.
In this vein, MIT SafePaths is in the process of building out both private tracing and a public health solution, part of which will utilize the new Google/Apple API, defined by two key elements: (1) open source software components (2) interoperable standards and backbone that work across GPS/BT/WiFi/Telecom
We will be using either on-device calculation or use encrypted trail match with guarantees of privacy. Thus it avoids the ‘big brother’ surveillance state problem in certain countries where GPS matching solutions have been very effective but are draconian. Such misinformation and distrust can cause civil unrest, especially in heterogeneous societies.
The vision at Safe Paths is to enable a fusion of various technologies including GPS/Bluetooth/WiFi SSID to provide a more reliable approach, reduce false positives/negatives, provide context for believability and also provide aggregate dashboard view for public health officials. SafePaths is already delivering these technologies and will continue to build them in an open source way. Follow pathcheck.org for a more technical discussion.