Procurement and early deployment of artificial intelligence tools for chest diagnostics in NHS services in England: a rapid, mixed method evaluation

Journal article

Published: 10/09/2025

Evidence suggests AI might benefit diagnostic services by supporting decision-making, improving detection accuracy, reducing errors, increasing efficiency, and easing workforce burdens. However, little is known about real-world implementation, including procurement, preparation for deployment, experiences of staff, patients, and carers, and impact on effectiveness and costs – due in part to very few studies having been conducted on real-world implementation of AI tools for radiology diagnostics.

In July 2023, the Artificial Intelligence Diagnostic Fund (AIDF) was launched to support AI deployment for chest diagnostics, including lung cancer, across 12 imaging networks (bodies created to support innovation in and standardised use of imaging diagnostics across NHS services across their local regions) and 66 of the 124 acute NHS Trusts (NHS organisations that provide a range of healthcare, e.g. acute hospital care or mental health care, to their local communities) in England.

 Improving chest diagnostics and lung cancer detection are national priorities due to high rates of late-stage lung cancer diagnosis, and efficient chest diagnostics may be central to managing any future respiratory disease pandemic.

To help increase understanding of real-world implementation of AI tools in a high priority healthcare setting, we analysed procurement and preparation for deployment of AI tools as part of the AIDF programme. Our analysis was guided by the Non-adoption, abandonment, scale-up, spread, sustainability (NASSS) framework, which considers the social and technical factors that interact to shape planning, implementation, and uptake of technological innovations.

Our analysis addressed the following questions:

  • How were AI tools for chest diagnostics procured and deployed as part of AIDF
  • Which factors (e.g. context, technology, implementation processes, capacity to implement, and stakeholder characteristics) influenced procurement and early deployment of AI?
  • What are the lessons for future real-world implementation and evaluation of AI in
    diagnostics?

Read the full journal article
 

Journal article information

Abstract

Background

Artificial Intelligence (AI) may support accurate, efficient radiology diagnostics. However, little is known about implementing AI in clinical settings. In 2023, NHS England launched a programme funding 12 networks of 66 NHS Trusts to implement AI for chest diagnostics, including lung cancer.

Methods

This was a rapid evaluation (March–September 2024) of procurement and early deployment of AI for chest diagnostics at network (n = 10) and Trust (n = 6) levels. Researchers interviewed network teams, Trust staff, and AI suppliers (n = 51); observed planning, governance, and training activities (n = 57); and analysed relevant documents (n = 166). Thematic analysis was guided by the NASSS framework.

Findings

Procurement and deployment of AI took longer than anticipated. Procurement involved engaging selection panels, assessing tenders, and contracting suppliers. Preparation for deployment required AI integration, governance processes, staff engagement and training, planning patient engagement, and collating impact data. Patient communication plans varied and were still developing.

Key challenges included: engaging overstretched clinical staff; limited AI knowledge; staff concerns about appropriate tool use; managing unsuccessful suppliers’ responses; and variation in local governance, IT systems, and data quality.

Enablers included: strong programme leadership, networks sharing expertise and capacity, committed clinical/technical/procurement specialists and suppliers, clinical champions, and dedicated project management.

Interpretation

Implementing AI is a complex socio-technical process requiring significant resources. Future implementation may be strengthened by allowing sufficient time and capacity, engaging stakeholders at multiple levels, and more explicitly considering patients and equity, diversity, and inclusion. Many of the influential factors mirror those seen with other healthcare innovations, suggesting that AI may not solve service challenges as straightforwardly as policymakers expect.

 

Read the full journal article