A reality check for AI tools in the NHS

Radiologist Dr Farzana Rahman examines the challenges for the adoption of AI tools in the NHS and why other critical issues must be addressed before such technologies can be adopted

Coming into 2024, we are truly in the midst of another AI hype cycle. As radiologists, my colleagues and I have become accustomed to claims that AI can solve the NHS’s workforce crisis. The truth is that this technology is incredibly exciting, but do we need a reality check before the NHS gets too carried away by the promise of AI?

The promise of AI

In recent years, we have seen huge excitement about the potential for AI to replace radiologists, particularly in the context of the radiology workforce crisis. The Royal College of Radiologists (RCR) has been calling for greater investment to grow the radiologist workforce for years, but these calls have been mostly brushed aside. The radiologist workforce grew by just 3% in 2022, whilst the RCR estimates that there is currently a 29% shortfall of clinical radiologists in the UK. ⁽¹⁾

Attempts to increase the consultant radiologist workforce are underway, but improvements are simply not keeping up with demand. Last year, 44 million imaging tests were reported in England alone – an increase of 26% from the previous year. This huge pressure on radiology services not only means that patients are waiting longer for their scan results but also that there is an increased risk of reporting errors. Only 24% of clinical directors currently think that their radiology department has enough clinical radiologists to provide safe and effective patient care. ⁽²⁾

AI tools in the NHS for medical imaging are even more attractive because healthcare systems simply cannot function without radiology. Over 3.6 billion radiology exams are conducted globally each year, including 43 million in the UK alone. Every medical department relies on radiology to provide diagnoses; without a diagnosis, treatment is often impossible.

We must also remember that AI technology isn’t just an innovation for higher-income countries but could also be used to address the huge inequalities when accessing healthcare resources worldwide. This is perhaps where there could be most impact for patients. A lot of the technology and skills are concentrated in a small number of countries, but AI tools that help do things at scale will have a hugely important role in improving global access to healthcare.

AI versus digital transformation

In June 2023, the UK Government announced a £21m investment in AI tools in the NHS, with Trusts able to apply for funding through the AI Diagnostic Fund to help speed up diagnoses. This was incredibly welcome news to those working in chronically overstretched radiology departments, but was this putting the cart before the horse?

Unfortunately, too many radiology departments still use paper-based or telephone-based systems, relying on siloed PACs (picture archiving and communications systems) and Radiology Information Systems (RIS), making it extremely difficult to review and share scans. Some hospitals even struggle to get a reliable wireless internet connection. I can’t have been the only one to wonder whether digitising the NHS’s diagnostic services to make them AI-ready would have been a better use of government funding. Instead, we’re now in the slightly surreal position of having funding for AI tools in the NHS but not for the necessary digital infrastructure to support it.

There are also huge benefits associated with digitising diagnostic services beyond supporting new AI tools. A 2023 NHS pilot at North West Anglia NHS Foundation Trust (NWAFT), in collaboration with radiology company Hexarad, digitised on-call diagnostic services at two major hospitals, which resulted in increased clinical capacity and faster delivery of emergency diagnostic services.

Early results from that pilot suggest that implementing the digital platform reduced vetting time by 84.3% (30 minutes to 4.7 minutes) and streamlined the management pathway of more than 8800 patients, aiding fast and accurate diagnosis. The time saved is estimated to be between 12 and 14 hours per night, the equivalent of one extra clinician per shift in a ten-doctor department. To achieve this same outcome, a hospital would need to spend £365,000 per annum on locum agency fees.

Results like these demonstrate the huge benefits of digital transformation for NHS diagnostic services, while at the same time putting Trusts in a better position to embrace AI technology when it is ready for wider implementation.

Looking to the future of AI tools in the NHS

Like many radiologists, I see a future where radiology departments (not just radiologists) can work seamlessly with AI tools. A patient’s radiology journey starts from the moment their doctor decides they need a scan, but several logistical steps need to happen before the scan results are ready. We spend a lot of time and resources on the most complex part of the pathway, namely interpreting the scan. However, AI and automation could streamline pathways such as patient booking or managing the supply and demand of both scanning and radiologist capacity.

However, the promise of AI shouldn’t close our eyes to the reality of the state of radiology in the NHS right now. In many places, the digital infrastructure isn’t there to allow AI tools in the NHS to work. We need to make sure that the solutions being developed and funded truly address our most pressing problems. Sometimes, those problems will be diagnostic, but often, streamlining workflow or data collection can be just as effective. Technology cannot fix a broken process.

It feels like we’re on another upward curve in AI at the moment, where the enthusiasm and excitement at the prospect of an ‘easy fix’ for the NHS are at an all-time high. Recent discussions about AI have done little to temper this in the popular imagination. Still, we must hope that NHS leadership can see that digital transformation, including AI, is what will make the biggest impact on improving diagnostic services.

Innovation very rarely follows a constant upward trend. There will certainly be a future for AI in healthcare, but it might not be quite the one we expect.

References

1. RCR Clinical Radiology Workforce Census 2022
2. NHS England Diagnostic Imaging Dataset Annual Statistical Release 2021/22