Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

NDORMS researchers have conducted a review of the literature on the use of AI in clinical decision-making. Published in the Lancet the study identifies why AI has not yet been taken up more widely, and provides a framework to better inform future adoption and integration of the technology.

AI used in medicine

Diagnosis is a cornerstone of clinical decision-making. Worldwide, there is increasing demand for diagnostic services such as blood tests or radiology assessments. However, this demand often strains healthcare systems already facing workforce shortages. Artificial intelligence has emerged as a promising solution to enhance diagnostic efficiency, accuracy, and accessibility and significant funds and resources are being put into AI from both governments and private companies. But despite this, the actual impact that AI has had on people's health care is still limited.

Rachel Kuo, NIHR Doctoral Research Fellow in Plastic Surgery at NDORMS is the lead author of a new study to understand the factors at play.

'Many research studies have shown us that AI can perform well in diagnosis, at least in computer experiments, and partly there is an element of waiting for the technology to get embedded,' explained Rachel. 'But it was clear to us that uptake of AI goes beyond performance measures. We wanted to explore the views of stakeholders involved in the development and implementation of diagnostic AI, and by understanding their perspectives we can better inform the adoption and integration of these technologies into healthcare systems. We also wanted to make sure that our research group had representation from patients and members of the public, so at the beginning five patient and public involvement (PPI) co-producers joined the team.'

To do this, the team conducted an extensive review of existing literature, analysing qualitative and mixed-methods studies focused on stakeholder perspectives on diagnostic AI. They screened 16,577 articles, of which 44 were included in the analysis. Within this, 689 participants had been interviewed and 402 included in focus groups sharing their views on diagnostic AI and its potential impact.

Published in The Lancet eClinicalMedicine, the study identified four distinct stakeholder groups: patients, clinicians, researchers, and healthcare leaders. One of the notable findings was an under-representation of patients, researchers, and leaders across the articles, despite them being a really important part of the decision-making process of whether to adopt the technology or not. The stakeholder groups were shown to have different priorities and key concerns regarding diagnostic AI, whether that be around privacy, security, trust, or benefits and weaknesses for patients.

To aid in the analysis and interpretation of the data, the research team modified the Non-adoption, Abandonment, Scale-up, Spread, Sustainability (NASSS) framework, tailoring it specifically to diagnostic AI. The framework provides valuable guidance for future research and implementation efforts, emphasising the importance of representing all stakeholder groups and addressing their priorities and concerns.

Rachel concluded: 'Further research is needed to delve deeper into the perspectives of under-represented stakeholder groups and to refine strategies for integrating diagnostic AI into clinical practice effectively. As the field of medical diagnostics continues to evolve, this study will help people thinking about using AI in research or implementing it in hospitals.'