I have attended the UKIO Conference for a number of years, but this year certainly feels different: 2023 feels like an inflection point for the role of technology in healthcare settings. I work for Qure.ai a company developing artificial intelligence for medical imaging, so you might expect me to be positive about the potential to improve diagnostics and treatment pathways. However, the overall theme for this year’s UKCIO is very appropriate: “Synergy and symbiosis: Breaking down barriers in healthcare.”
In the past, there has been a clear distinction between technologists and clinicians at UKIO, but now there appears to be more collaboration between these groups, which is important. Technology is having a big impact on how services are delivered in many industry sectors and it is no different in healthcare, particularly when it comes to AI. AI is not a silver bullet to solve the many challenges facing the NHS, but it can be an enabler and trusted partner to healthcare professionals.
Why am I convinced this partnership between technologists and clinicians can deliver significant benefits to healthcare professionals – and more importantly, patients? A parallel example is the use of autopilot systems in aircraft. Travellers are accustomed to this, but there is always the reassurance of a human operator, a pilot, to oversee decisions made by the computer system.
Adopting AI in healthcare settings should be seen in a similar light. There is a lot of hype about AI systems being sentient and some quite ludicrous claims about machines taking over. This is not the vision for AI that we have at Qure.ai. We are already working with 25 NHS Trusts in the UK, more than any other provider of AI for medical imaging, and it is clear our technology works best in a symbiotic relationship with the consultant radiologist or radiographer. We have seen it deliver significant benefits such as identifying incidental findings and automating the monitoring of disease progression. We can detect mislabelled findings and we are one of the few providers who can attribute ‘normal’ to an X-Ray.
With the shortage of trained radiologists and radiographers, AI can automate initial analysis to ease workloads. We have seen this work well in emergency settings where nightshifts are short-staffed and we have also completed academic research recently demonstrating that AI can support healthcare systems where access to radiologists is limited. Ultimately, this is about helping to clear backlogs and improve the prospects for patients. Our technology is being used to detect lung cancer and we all know how crucial it is to spot abnormalities earlier for treatment to be successful.
Getting to the point where these benefits are realised is not straight forward, which is why we looking forward to running a session with clinicians at UKIO to discuss how we put the fundamental building blocks in place to ensure ethical and safe adoption of AI in the NHS: “AI in clinical practice – How to get past the hype.” The UK Government is doing a lot to encourage the development of AI solutions, which is being matched by the efforts of various NHS bodies to create the right conditions for its adoption.
We welcome the Medicines and Healthcare products Regulatory Agency (MHRA) seeking to implement its programme to encourage the adoption of innovative new medicines and medical technologies. This is positive but we must ensure there is agreement on a standards framework to benchmark these innovations. Such standards must address key questions such as data interoperability, access to scientific evidence and application of regulations.
Safe and secure connections between NHS IT systems and external providers will allow different medical applications to work in tandem to interrogate patient data. This is not a request for open season on patient data, but greater interoperability between applications will create a more holistic picture of a patient’s medical history for more accurate diagnoses. While we can demonstrate the effectiveness of our AI based on our research, clinicians expect us to show how we can integrate data to create an even more accurate understanding of local demands.
Peer-reviewed scientific evidence is also vital in giving clinicians confidence, because the AI decision making process must be explainable and transparent. If a healthcare professional cannot understand why the system has made a certain decision, that will limit progress. Regulatory approval is also essential, but it must evolve to keep up with the pace of technological change.
As we meet in Liverpool, I hope collaboration will be top of everyone’s agenda. Clinicians must build business cases for AI which requires a new set of business and multi-disciplinary skills. We should also be creating an ecosystem for technologies to work together to automate and improve a variety of operational back office and diagnostic processes. This will require partnership between everyone working in the healthcare system, so breaking down barriers and encouraging collaboration are going to become even more important.
By Darren Stephens, SVP & Commercial Head UK and Europe, Qure.ai