Short Abstract

This presentation will share how a participatory approach was combined with AI-assisted (LLM) review to map the networks and relationships supporting Suffolk’s physical activity participation. We will highlight learning from integrating these methods to understand a complex system.

Description

Suffolk, a county in the East of England, experiences consistently low participation in physical activity and marked health inequalities. Traditional, individually focused or programme-based approaches have not been sufficient to achieve sustainable change. Stakeholders recognise that meaningful, long-term improvement requires coordinated action across multiple organisations, sectors, and settings. This has led to a shared commitment to a whole-systems approach and an increased focus on understanding how the system operates, who is involved and where there are opportunities to work together differently.

This presentation will outline how a participatory approach and AI-assisted review were combined to map networks and relationships that support participation in physical activity across Suffolk. The work builds on Suffolk’s ongoing whole-systems approach to physical activity, where earlier evaluation highlighted the need to understand who is currently involved in the system, how organisations connect and where opportunities exist to broaden engagement across sectors.

The project used social network analysis to map relationships and connections between stakeholders across local government, health, voluntary and community partners. An initial participatory approach, based on four structured questions, was used to create an early network map in Kumu (TM). AI-assisted review was then used to examine grey literature and add further detail, helping identify information missing from the initial map. The final and ongoing stage involves validating the map with partners to check that it accurately reflects the space they are working within.

Integrating these methods supports a participatory approach that benefits from the expertise and perspectives of stakeholders within the system, while also reducing the capacity required to develop and update the map. This is important given that systems, and the organisations within them, are dynamic and their relationships non-linear and unstable. Bringing these elements together can help build network maps more efficiently and support their ongoing revision, allowing them to remain current and become living documents that are actively used across the system to guide action, reflection and engagement. They may also support discussions about who is well-connected and able to support change, and where important voices or organisations are not yet represented.

The session will outline the steps taken to design the approach, the practical benefits and limitations of combining social network analysis with AI-assisted review and how this supported shared understanding across the system. A visual map is being iteratively developed with partners and will be shared during the session alongside emerging learning.