Description of icon when needed 11 Min Read

By Suzanne Ambiel with Adam Fowler

It’s been a year since the global onset of COVID and the subsequent and persistent community lockdowns to help prevent spread. While a blunt instrument, lockdowns and restrictions in gathering and social settings have proven effective in tamping down the spread of COVID. But their impact devastates local economies and social structures. Instead of imposing severe restrictions on an entire community or economy, what if you could be more precise – specifically identifying those who may have been exposed and quarantining just those individuals, instead of the entire community? This question proved to be the spark for Adam Fowler and the recently released open source project, Herald. 

Herald is an open source Bluetooth protocol, intended to improve the accuracy and efficacy of contact tracing. Conceived to help address the immediate needs of the COVID pandemic, the project’s maintainers envision other uses for the technology, including medical equipment exposure tracing and food supply safety. 

The release of Herald on GitHub in late fall of 2020 and its recent acceptance into the Linux Foundation – Public Health (LFPH) has unleashed a flurry of activity for its primary maintainer, Adam Fowler. He’s been tough to catch ever since, but the Open Source team finally pinned him down for a quick chat about Herald – how it started, where it’s headed and how you can get involved. According to Adam, Herald started with the seeds of an idea on a train ride to London and today is a globally accessible open source project adopted by regional and national entities. All in less than a year. With goals of using the newly formulated Herald protocols for ubiquitous deployment in the measurement of exposure for not only people but also for sterile products or equipment, Adam takes us through its recent and rich history.

Phase One: Covid-19 Lockdown

March 2020 marked the beginning of the lockdown, providing the general public protection from COVID-19 in the United Kingdom. There was an immediate need for sophisticated contact tracing to help prevent infection. What better way to manage this than through the use of technology everyone already carries in their pockets: mobile phones. The solution seemed deceptively simple in measuring the proximity of each new case in order to alert those who may have been exposed by proximity based on the position of their mobile devices. With that in mind, Adam began work on his first Proof of Concept while on a train to London.  

Within a week, Adam and a team of VMware engineers built a demo, and had the code working in a limited mobile application. This application gave five bands of distance rather than a measure in meters. The team realized that for the epidemiological models to be accurate, this was not sufficient. The solution required a more precise distance conversion.

The team then ran into another challenge. For mobile phone-based contact tracing to be accurate, the technology needed to work with and on as many types of mobile devices as possible. In this situation, more is better – but the unpredictable and diverse set of possible devices and versions meant the team needed to think differently. 

Phase Two: In Contact Tracing, More is Better 

When Adam was working as the Lead Architect for another contact tracing application, problems were identified in terms of the quantity of different devices. Working with epidemiologists and other experts solved many of these problems, but others remained. After handing over the contact tracing app he was working on, Adam spent his spare time considering the core challenges. The wide range of Bluetooth tolerances and implementation-specific variations in those devices makes the task of accurate and reliable digital contact tracing very difficult. The sheer variety meant the team needed to seek a common, shared standard. But as Adam phrased it, “The problem with standards, is that everyone uses their own.”  

Adam decided to go back to the drawing board and determine what technology factors affect epidemiological accuracy for COVID-19 risk. Reading a wide range of scientific papers, especially Oxford University Big Data Institute’s seminal paper on COVID-19 and Digital Contact Tracing, Adam devised the Fair Efficacy Formula and published this as a paper. “The problem was there was no standard way of testing efficacy for any contact tracing app, and no way to link how devices were used against the required epidemiological effects of tracing and isolation.” The Fair Efficacy Formula provides this standard way of comparing any digital contact tracing app’s efficacy.

The Fair Efficacy formula calculates the ability of any proximity detection protocol to accurately detect, measure and therefore assess the epidemiological exposure risk incurred by a mobile phone owner. These results can then be plugged into Oxford BDI’s OpenABM based COVID-19 spread simulator to gauge an application’s control effect on the spread of a virus in a population. To the team’s knowledge, Herald is the only DCT system shown in simulations to have a halting effect on COVID-19 on its own, with no other control interventions.

Creating Herald

The Fair Efficacy Formula provided valuable insights for the team. Pulling back to look at the science, the team devised a new approach to solving for accurate contact tracing using mobile phones and the first versions of Herald emerged five weeks later. The project tested the mathematics of the efficacy of real-life contact tracing. Since the initial release, continuous iteration led to improvements in power consumption and removal of interference. 

Today, Herald provides a protocol for measuring applications against a standard set of tests, enabling more accurate contact tracing applications. The Australian Government’s CovidSafe application, and the recently relaunched Alberta Trace Together application, are among applications that were developed in cooperation with the Herald community contributors. Now that the protocol is verified, the community envisions additional  potential use cases including food safety in supply chains and other relevant public safety solutions. 

“The ultimate goal is to make digital contact tracing so reliable that it negates the need for geographical lockdowns.”

Open Sourcing it for All

Democratizing access to the information creates a sense of fair play for everyone in the world and invites the best and brightest to build solutions. Founded in summer 2020, the initial focus of LPFH has been helping to deploy applications implementing the GAEN system. The organization has been growing into other areas of public health that can take advantage of open source innovation. Working closely with LFPH and its contributors, along with beneficiaries of  the technology such as FaceDrive, TraceTogether and developers within the Ministry of Health for the Australian government, the Herald project continues to expand its reach. 

Every use case and country government will use their own methodology for contact tracing. The Herald team can provide the initial research to lower the risk of these unique approaches, centralized or decentralized. For example, Herald was ported from Android/IoS to C++  to enable use in  embedded devices and wearables (fitness or health trackers) , which opens the protocol for usage across a range of applications. Supporting detection of devices running other protocols such as GAEN and OpenTrace and creating a Herald interoperability standard helps with cross-border or state line traceability management, and benefits all countries no matter which protocol their application uses. 

What’s Next for Herald?

The future looks bright for Herald. According to Adam, Europe’s ETSI standards body asked him to write the Herald Interoperability approach into their upcoming standard on DCT interoperability. In the coming months, the Herald team will be building functionality to provide on-device data analysis for risk estimation amongst your own contacts. Additionally, the team plans to release APIs for various distance conversion and risk estimation models, and to look beyond COVID-19 to Ebola and other diseases. 

At over 2.3 million device data points, Herald already provides the scientific community with reference data for broad based studies on epidemiological risk models. The ultimate goal is to make digital contact tracing so reliable it will eliminate the need for future COVID or other contagion-related geographic or community lockdowns. 

Getting Involved with Herald

Nearly everyone can help contribute to this project. It’s not necessary to be a developer – the Herald community needs help in a variety of ways from data science, viral behavior, security analysis and ethics to documentation. Join the Herald team to help build contact tracing accuracy. You can find more at the Herald Project Page.