Showcasing our award-winning PETs solution at The Royal Society
Andrew presented our location-based approach to Privacy Enhancing Technologies (PETs) at the Centre for Data Ethics and Innovation’s PETs Demo Day, on 22 May. The event celebrated the US-UK PETs Prize Challenge winning solutions. We wrote about our award for meaningful consideration of data governance, here.
At the Demo Day, Andrew introduced Diagonal’s award-winning White Paper (which you can read here). The showcase was hosted at The Royal Society, which included presentations from all the prize winners.
The challenge was to use PETs and federated learning models to inform pandemic response planning, using simulated data. Andrew shared how we used a place-based approach, and an explainable approach in our solution. In our model, we assigned places, rather than people, with infection risk scores. This score was determined as a function of the total number of visitors compared with the infected visitors.
While this challenge was not initially framed as a city-planning challenge, we saw a strong link between it and our purpose: to improve the quality of life of those living in cities (as stated in our charter). This challenge appealed to us because it allowed us to explore the opportunity for privacy-preserving technology in place-based models. The approach we used could be applied to other questions that come up in city planning, like: how are e-scooters being used in a city, and by whom? How do local communities use – or not use – public amenities? How might healthcare services and local authorities coordinate to make care more accessible to patients?
To design inclusive, equitable city spaces we need data about how people use spaces, which communities use – or don’t use – services, and how people’s needs vary by district. However, this data is very sensitive. Sharing this data for analysis should be governed in a considered, explainable and transparent manner. We talked about the challenges of anonymising location-based data in an earlier post, here.
At the Demo Day, Andrew explained how our approach to sharing sensitive data preserves the privacy of data subjects, and how this approach can be applied to multiple use cases. The challenge was to help predict the spread of infectious disease. Our approach was specifically designed to help to combat concerns around public trust. Government and organisational trustworthiness is key consideration when assessing the use of personal and potentially sensitive data. The method we developed offers transparency upon which public trust can be built and maintained, allowing for social approval when using sensitive data. This is crucial in the domain of public health, where relying on individual consent can make large modelling projects problematic or complex. But it is also very relevant in city planning and shaping the built environment.
Diagonal’s presentation was part of the Demo Day’s focus on ‘bolstering pandemic capability responses’.
The event included discussions around tech advances with stakeholders from government, the finance and healthcare sectors, representatives from industries adopting PETs, international agencies, privacy specialists and investors.
The Royal Society’s Privacy Enhancing Technologies programme in 2023 included this report, From privacy to partnership: the role of Privacy Enhancing Technologies in data governance and collaborative analysis (PDF), created in collaboration with the Alan Turing Institute.
To talk more about our White Paper, our work in PETs or the new tooling we are building please do get in touch here.
