Trustworthy and ethical assurance of digital twins: Putting the Gemini Principles into practice

The UK has invested heavily in digital twins (DTs) across both the public and private sectors, as well as a wide range of domains and use cases (e.g. transportation, infrastructure, health, energy, and natural environment).  

All these sectors and domains, however, face a common challenge: DTs are a relatively immature technology (at least outside of advanced manufacturing and engineering), and developers and practitioners increasingly make use of varied data-driven technologies, such as artificial intelligence, throughout all stages of a DT’s design, development, and deployment lifecycle. The novelty and potential complexity of such data-driven technologies raises an important question:  

Which processes should developers follow throughout the lifecycle of these technologies to ensure that the DT is trustworthy and ethical? 

If the significance of this question isn’t immediately apparent, consider the following: 

• DTs are being used in healthcare to model complex traffic of activity across genetic regulatory networks, the passage of a breath through airways, the structure and function of the human heart, and there are even early attempts to model a virtual brain.² 
• DTs are expected to support decision-making in safety-critical national infrastructure (e.g. national air traffic services) to help build crisis resilience.³ 
• DTs are being used to help researchers build detailed maps of the natural environment, using data from across multiple spatiotemporal scales, to better understand the risks posed by climate hazards and identify optimal strategies for intervening and mitigating these risks⁴. 

In short, DTs are no longer just a way for organisations to eke out small efficiencies or cost-savings to help gain a competitive market advantage—although this is frequently touted as a benefit. They are fast becoming useful and usable tools for science and innovation to help address some of our most significant societal challenges.  

This goal will only be realised if developers of DTs can provide robust and clear assurance that they have been designed, developed, deployed, and maintained in a way that is trustworthy and ethical.

To support this goal, the DT Hub collaborated with the Alan Turing Institute (and others) on a scoping project known as Trustworthy and Ethical Assurance of Digital Twins (TEA-DT).⁵ 

Trustworthy and Ethical Assurance of Digital twins

The scope of the TEA-DT project was wide-ranging, but three objectives are important here:

1. to develop an open-source platform, known as the TEA platform, for building clear and accessible forms of assurance regarding desirable goals of a digital twin;
2. show how the Gemini Principles can be operationalised for the purpose of trustworthy and ethical assurance; and
3. adopt a participatory approach to assurance, grounded in real-world use cases of digital twins across the domains of health, natural environment, and infrastructure.

Focusing just on the first goal, the TEA platform has been designed to help DT teams (and their host organisations) show how specific goals or objectives, including the Gemini Principles, have been realised within the context of their DT project, and to communicate this clearly with relevant stakeholders and users.

This is achieved by developing a structured argument, known as an assurance case, that takes a top-level goal and justifies its validity by reference to a series of specific claims about properties of their project or DT, which are linked to clear (and where necessary auditable) pieces of evidence. This is important because a claim without evidence is unfounded; but evidence with claims is unexplained.⁶

Screenshot taken from the TEA platform, showing an assurance case for one of the Gemini Principles (Quality)

Project Report

To read more about the other objectives and see how the TEA platform was used to support a participatory approach to assurance, you can read the project report which is available for free here:

The report covers:

• An overview of the TEA-DT project
• An introduction to the TEA platform
• An introduction to the Gemini Principles
• Co-designing the assurance ecosystem
• Description and analysis of a community survey
• Two featured project case studies from the British Antarctic Survey and Networks of Cardiovascular Digital Twins project 
• PLUS Open challenge and next steps

Citation Information:

Burr, C., Anderson, J., Arana, S., Block, D., Byrne, J., Goodman, R., Gavidia-Calderon, C., Habli, I., Moreira, N., Niederer, S., Polo, N., & Wagg, D. (2024). Trustworthy and Ethical Assurance of Digital Twins: Putting the Gemini Principles into Practice. Alan Turing Institute. https://doi.org/10.5281/zenodo.14216050  

[1] With thanks also to Karen de Cesare, Catherine Condie, Ryan Goodman, Nury Moreira and the rest of the TEA-DT project team who helped plan and deliver the events discussed in this post.

[2] Coveney, P. and Highfield, R. (2023) Virtual You: How Building Your Digital Twin Will Revolutionize Medicine and Change Your Life.

[3] UKRI (2024) Core research challenges in digital twinning for crisis resilience. https://www.ukri.org/opportunity/core-research-challenges-in-digital-twinning-for-crisis-resilience/

[4] Data61/CSIRO (2024) Spark. https://research.csiro.au/spark/

[5] The TEA-DT project is led by the Turing Research and Innovation Cluster in Digital Twins (Alan Turing Institutes) in collaboration with the Centre for Assuring Autonomy (University of York), the Responsible Technology Adoption Unity (Department for Science, Innovation, and Technology) and the Digital Twin Hub (Connected Places Catapult). The project was generously supported by the UKRI’s BRAID programme, with funding awarded to Dr Christopher Burr. More information can be found here: https://www.turing.ac.uk/research/research-projects/trustworthy-and-ethical-assurance-digital-twins-tea-dt

[6] Kelly, T. P. (1998). Arguing Safety – A Systematic Approach to Managing Safety Cases [University of York]. https://www-users.cs.york.ac.uk/tpk/tpkthesis.pdf