Human-centric digital twin

AquaSavvy straddles the spectrum from research to early implementation.

Technically speaking, a more human-centric digital twin becomes possible through our approach of systems-of-systems integration of data sources (e.g., environmental, terrestrial, meteorological, and marine) at multiple levels (i.e., syntax, semantics, or conceptual), including those in different sectors e.g., the private sector or government.

Rather than integrating each system on a one to one basis, this project achieves interoperability between water-related data sources through an Evolutionary Architecture described by IPSME (Nevelsteen & Wehlou, 2021). Instead of only gathering diverse stakeholders around a table in an attempt to reach a common approach to data management, IPSME is conducive to rapid prototyping and iterative development. In the Evolutionary Architecture, Stakeholders can be added and removed dynamically to the ecosystem, in runtime, and with no downtime for the systems being integrated.

Rather than re-engineering existing data sources, IPSME supports legacy systems i.e., where the technical knowledge of the system as been lost or the risk of introducing errors in long running systems would be detrimental.

Artificial Intelligence (AI) will be employed to implement interoperability in IPSME and assist in analysing and manipulating data, external to the systems being integrated for pattern recognition in decision making or different visualizations of the data. Machine learning models, colloquially known as “cooperative AI” can assist decision-making in the domain of water (Wang et.al., 2024).

The function of the language model-based AI is to serve as a knowledge base and reduce misunderstanding among stakeholders. It is useful especially when complex tasks need to be decomposed into smaller pieces. Multi-agent coordination algorithms can then be employed to provide comprehensive model outputs.

In parallel to the technical tool development, participatory planning initiatives will be employed and compared based on different approaches being developed by the consortium members, allowing co-learning and contextual insights:

• actor landscapes as visual canvas (Quizau & Hoffmann 2023),
• incorporating creativity in the scenario development process (Gökmen & McKiernan, 2025) and
• adaptive planning (Adolphson & Bendz, 2025).