ARC NWC PhD student Alexander d’Elia, based at the University of Liverpool, is reviewing sociotechnical analysis and stakeholder exploration of how to make artificial intelligence (AI) a force for health equity in English primary care.

The project addresses health inequity and the implementation of artificial intelligence (AI) in primary care. Health inequity is the central subject of the new Governmental Office for Health Improvement and Disparities and part of the NHS Long-Term Plan. as well as one of ARC NWCs founding principles.

Alexander said: “The project sets a methodological precedent on how complex interventions can be assessed “prospectively” from a sociotechnical perspective. Ultimately, the service users of UK primary care will benefit, in particular those with the largest needs and those belonging to socioeconomically disadvantaged populations.”

Highlighting the importance of his project Alexander continued: “AI in primary care holds great potential, the difference with AI from other innovations is the pace of innovation. Previous waves of innovations have happened at a more gradual pace, allowing for a more controlled implementation.

The project carves out a niche against the preceding research. Namely, in contrast to preceding works as it takes a systematic, empirical approach specifically focused on the implementation setting that is English NHS primary care.”

Alexander is am aiming to map the ecosystem involved in the implementation of AI in English primary care, and from a sociotechnical perspective assess how this network of actors can be conducive to improving HI.

Alexander added: “The project has created an empirically grounded set of recommendations for how AI can be implemented in an equitable manner especially in Local Trust setting and NHS England typically leaves clinicians and patients behind. All stakeholders need to be on board for implementation success as it has been proven that innovation.”

AI is the subject of much regional, national and worldwide interest and there is a wave of funding calls and research coming out on how to best utilise AI in healthcare.

Alexander works along the Care and Health Informatics theme. He added: “Artificial Intelligence (AI)-augmented interventions are currently being rolled out across primary care, but the sociotechnical theory for deploying AI is in its infancy.

The current literature focuses predominantly on reducing health inequity (HI) by minimising algorithmic bias. Applying AI in healthcare will affect HI beyond algorithmic bias, through interactions with existing societal health inequities. There is a need to understand how the ecosystem in which AI is being implemented can be made to benefit HE through AI.”

Alexander has consulted with two ARC NWC public advisers in the development of the PhD project and in conducting a systematic review. Service users were involved as research subject (four focus groups of people living with diabetes), being interviewed about their thoughts on AI and primary care.

“So far, I have conducted a systematic scoping review which was followed by an ethnographically anchored inquiry based on 32 interviews with stakeholders including commissioners, decision makers, AI developers, researchers, GPs and patient groups.

This was complemented by an analysis of UK primary care data to assess the risk of algorithmic bias in big-data applications such as AI systems.”

Alexander highlighted the key results of the project: “AI is likely to impact HI in primary care through a multitude of mechanisms, including both those intrinsic to the AI systems (e.g. algorithmic bias) and wider system- and societal impact.”

“The research has found that Regulation and policy cannot guarantee equitable implementation of AI, but needs to provide a baseline framework to enable other stakeholders to promoting equity in the implementation process.”

d’Elia A, Gabbay M, Rodgers S, et al Artificial intelligence and health inequities in primary care: a systematic scoping review and framework Family Medicine and Community Health 2022;10:e001670. doi: 10.1136/fmch-2022-001670