AI’s promise in healthcare was a beacon of hope to improve patient outcomes and reduce pressure on overworked medical staff.
However, a new UK survey reveals a more complicated reality. Implementing AI tools at NHS hospitals is a much bigger challenge than initially expected.
The study, led by researchers at University College London (UCL), found that the journey from the concept of AI to clinical reality plagued by governance issues, lack of staff training, and difficult hurdles to integrate existing NHS IT systems with new technologies.
The findings serve as an important learning opportunity for policymakers, including the UK government. The UK government has identified digital transformation and AI as key pillars of the NHS long-term plan.
Pioneering Programs and their Unexpected Delays
The research, funded by the National Institute of Health and Therapy (NIHR), focuses on the NHS England programme, which was launched in 2023.
The initiative aims to introduce AI tools to help diagnose chest conditions, including lung cancer, across 66 hospital trusts.
Based on the £21 million funding, the initiative was designed to enhance diagnostic services by prioritizing key cases to review scans and flagging anomalies.
Previous laboratory-based studies suggested that AI could benefit diagnostic services significantly by improving accuracy and reducing errors, but this new study provides one of the first real-world analyses of AI implementation in healthcare settings.
The UCL-led team, which also includes experts from Nuffield Trust and Cambridge University, delved deep into the procurement and early deployment of these AI tools.
Through interviews with hospital staff and AI suppliers, they uncovered both the pitfalls and practices that help smooth the process.
The findings showed that the deployment of AI tools took much longer than expected.
Contracts alone were delayed by 4 to 10 months, and 18 months after the initial target of completion by June 2025, a third of hospital trusts (23 out of 66) had not yet used AI in clinical practice.
Human and technical hurdles for AI adoption
This study identified many important issues that slow implementation. The big problem was the extremely difficulties attracting clinical staff who were already working on very high workloads.
Many staff also lacked a basic understanding of new technologies and expressed some degree of skepticism about using AI in healthcare.
This was especially true for more senior staff who were concerned about accountability and the possibility that AI could make decisions without human supervision.
The technical challenges were equally difficult. New AI tools had to be embedded in the NHS’ aging and diverse IT systems. This is a complex task that differs from hospital to hospital.
The technical nature of the procurement process has also proven to be a hurdle. Some staff have found themselves overwhelmed by the vast amount of detailed information, increasing the likelihood that critical details will be overlooked.
Future paths: Best practices and recommendations
Despite the challenges, this study highlighted several factors that contribute to a smoother implementation.
Dedicated project management was an important success factor, as well as high level of commitment from hospital staff leading the implementation.
This study also found that shared learning and resources between local imaging networks, as well as strong national program leadership, can help drive the process.
In their conclusion, researchers said that AI tools can provide valuable support for diagnostic services, but “it may not be easy to address current healthcare services pressures as policymakers would like.”
The authors of this study have created several key recommendations to improve the future deployment of AI in healthcare.
They highlighted the need for early and continuous training for NHS staff on ways to use AI effectively and safely, and emphasized that this training must address concerns about accountability and clinical input.
They also suggested that creating a nationally approved finalist list of potential AI suppliers would help streamline the procurement process for individual hospital trusts.
Next steps in research
Researchers are currently conducting further research to understand how tools will be used once they are fully embedded and explore the patient and caregiver perspectives. This is an important aspect that was not part of this early stage.
This ongoing work is built on limited but growing evidence on real-world AI implementations, paving the way for more effective and successful integration of technology into the NHS future.
Source link
