Implementing AI innovation in radiology departments in the English NHS: a qualitative study on the experiences of professionals, patient groups and innovators

TL;DR

This study explores how AI tools for detecting lung cancer are perceived and implemented in the English NHS, highlighting challenges and perspectives from professionals and patients.

Contribution

The study provides new insights into the practical and contextual challenges of implementing AI in radiology within the NHS from multiple stakeholder perspectives.

Findings

AI is seen as a supportive tool rather than a replacement for professionals in diagnostic pathways.

Integration into workflows and perceived value strongly influence professionals' acceptance of AI tools.

Fragmented legal frameworks and unclear governance are major barriers to AI implementation in the NHS.

Abstract

Digital solutions and Artificial Intelligence (AI) innovations are often presented as the answer to many challenges faced by healthcare systems around the world. The UK government has made significant investments in this area, yet there have been concerns about the challenges faced when these technologies are implemented in practice. The aim of this study was to explore the perceptions and experiences of professionals, patient groups as well as innovators of introducing a new AI innovation used to detect potential abnormalities for lung cancer in radiology departments in the English NHS and to investigate the implementation challenges from their perspectives.

Between September 2022 and January 2024, we visited five sites and conducted 34 interviews with radiologists, radiographers and other professionals involved in the implementation of the tool. We also interviewed seven staff from the company who developed and implemented the tool. In addition, three 2-hour focus group workshops, two online and one in person, were conducted in January 2024 with a total of 14 patient and public involvement and engagement (PPIE) participants from diverse regions, backgrounds and lived experience across England. Following initial coding done inductively, the Consolidated Framework for Implementation Research (CFIR) was applied as an organising framework to structure and interpret the emerging themes.

Both professional and PPIE groups recognised the potential of AI in the diagnostic pathway, while generally seeing it as a supportive second pair of eyes rather than an autonomous decision-maker, particularly when delivering sensitive news and information. Professionals’ acceptance depended on how the tool was integrated into existing workflows and its perceived value, with triaging functionality seen as effective, but varying in usefulness depending on local workload pressures. Innovators as well healthcare professionals highlighted a number of implementation challenges, particularly around fragmented legal and regulatory frameworks and unclear governance within the NHS.

Our findings underscore that successful AI implementation in clinical practice depends not on the technology alone but on its effective integration into existing healthcare contexts and alignment with the beliefs and needs of healthcare professionals, patients and the public.

Figures