Scottish Health and Industry Partnership AI lead, JD Blackwood, discusses the vast potential of new technology to positively impact preventative efforts.
If the COVID-19 pandemic can be said to have had any kind of positive legacy, then it must surely lie in its role as a catalyst for wider collaborative change in healthcare – necessitating the rapid development and adoption of new solutions to improve operational and clinical outcomes.
Technical and clinical expertise is combining to tackle major health challenges, and Scotland – primarily through the work of the Scottish Health Industry Partnership (SHIP) – offers a rich test bed to work with industry to develop solutions that improve the quality, efficiency and sustainability of healthcare.
However, as we continue to wrestle with the pandemic, key areas require significant attention, if not more so given the huge level of NHS resources deployed in the fight against COVID-19. With cancer screening programmes paused, and reduced referrals from GPs and emergency departments during the pandemic, concerns of a post-COVID cancer crisis have been widely reported.
Those concerns have been further compounded by an accompanying shortage of oncologists, radiologists, and other healthcare professionals across the cancer diagnostic and treatment pathway. Indeed a third of radiologist posts remain unfilled – a situation that is likely to worsen while demand increases.
Cutting-edge innovation with artificial intelligence (AI) offers potential solutions to tackle delays and improve long-term outcomes in cancer screening programmes. AI represents nothing less than an opportunity to radically change health and social care – and quite simply we cannot recover from the pandemic using human resource alone.
The application of AI to solve health and social care problems is a young and rapidly developing field. It is important that we invest in the whole innovation lifecycle, including research and development, skills development, business growth and support, and robust evaluation within the clinical environment. It is not enough for technology to be interesting – it has to be viable and properly embedded into clinical practice. An investment in Scotland’s rich AI innovation ecosystem, will generate national economic prosperity and improve the health and wellbeing of the nation.
Within Scotland we already boast several immediate advantages for realising AI’s true potential through top-quality academic and commercial research capability as well as long-standing collaborative relationships between NHS, industry, and universities.
Successful programmes like iCAIRD demonstrate that with unparalleled, secure access to healthcare data, Scotland can invent and implement AI solutions at pace. Thanks to investments in Scotland’s data infrastructure, such as Research Data Scotland and the Safe Haven network, and the creation of Regional Innovation Test Beds, the NHS is working hand-in-glove with innovators to robustly validate technology.
In terms of adoption, the Centre for Sustainable Delivery (CfSD) is taking innovation and accelerating its integration into clinical practice, providing clear signals to industry about health and social care needs so they can develop appropriate solutions.
If we focus on cancer screening, and consider the application of AI in offering a more efficient approach then the potential is significant.
Cancer screening programmes are designed not only to detect signs of the disease, but also to detect risk indicators and the absence of cancer. They help the NHS prioritise and focus patient care on those with the greatest need, at the time they most need it. Patients are regularly tested and in most cases, no cancer will be detected. Highly trained medical professionals have to visually identify signs of cancer, some of which are difficult to spot.
A large amount of clinician time is being spent on tests where no cancer is detected. For example, only about 2.8% of bowel cancer screening tests are positive for cancer risk indicators. AI excels at this type of task, doesn’t experience fatigue like humans do amid mass screening, and bases its analysis on rigorous, comprehensive data and pattern recognition.
Automating this first step of the process, which is then validated by clinicians, reduces the pressure on resources, the impact of delays in staff recruitment, and the shortfall in skilled clinicians. As a result, more effort can be focussed on the positive cases, and clinicians can process more of them, including cancers that are rare or difficult to diagnose. Ultimately, patients benefit through faster diagnosis and treatment and the NHS benefits through greater efficiency.
AI is also able to spot more subtle cancers, or to better and more quickly identify cancer risks, by combining information from multiple tests and a patient’s broader medical history. The automation of cancer detection and grading using AI will in the near future provide patients with a more precise and timely diagnosis and increase the chances of successful treatment.
Identifying such cancer risks, and spotting subtle changes that can sometimes go undetected by the human eye could be transformational for healthcare. With that in mind, we already have evidence of AI being successfully deployed to learn from existing data, identify issues, and make connections in everything from lung cancer cases, to predicting the risk of ovarian cancer, and aiding early diagnosis of oesophageal cancer.
Impressively, AI can detect signs of very early cancers but also identify changes to those cancers’ DNA and where exactly in the body they are developing. It is thought to be capable of picking up signs of more than 50 types of cancer. The Galleri test uses a machine learning algorithm to examine cell-free DNA (cfDNA) that leaks from tumours into the bloodstream. It detects chemical changes, known as methylation patterns, with a false positive rate of only 0.5%, and the technology is being piloted in NHS England with the next stages of the study being considered for Scotland.
Offering a clinical viewpoint, Professor Colin Fleming, consultant dermatologist at NHS Tayside who is leading the AI Skin Cancer Consortium in Scotland, said: “AI does not immediately put all the answers at our fingertips, but it does offer huge potential in future if we can get the balance right. We need to invest in it first and aim to devise systems that are sustainable.
“In my area of dermatology, one in five GP consultations arise from skin complaints. Feasibly, AI could really help with diagnosis by directing patients quickly to pharmacists for their particular complaint and reducing waiting lists. The key question though is, ‘How do we do that well?’ That’s where AI benefits must be carefully weighed against potential drawbacks and rigorously tested before being deployed.”
“Nevertheless, the possibilities remain vast and particularly so if that successful balance can be achieved.”
Beyond the potential benefits for healthcare that we are already seeing in dermatology, radiology, cardiology, pulmonary disease, and diabetic retinopathy, AI is also a growing market with opportunities for Scottish spin outs and start-ups to expand and dominate the international scene.
Scotland is well placed to attract investment, accelerate talent growth, and become a go to location for accessing robust medical data. Our infrastructure continues to develop, with AI a firm part of the NHS Scotland recovery plan, Scotland’s digital health and care strategy, Scotland’s AI Strategy, and more widely, the UK Life Sciences Vision.
As part of that, the Scottish Government is committed to delivering a national life sciences, health and care AI Hub, providing a single place where people can access advice, infrastructure for AI development, standards, investment, and growth. The hub will be a cornerstone element for delivering the life sciences element of Scotland’s AI strategy which aims to pull all of the sectors together to build a national capability.
An important part of the AI ecosystem, and at the forefront of this work, is SHIP – a government team catalysing innovation within the Chief Scientist Office and leading on the development of a demand signalling plan outlining priority areas for innovation, as well as the development of AI capacity and capability within Scotland.
SHIP’s role is complemented and underlined by close working with Living Lab, BioQuarter, the Moray Growth Deal, Innovation Test Beds, the Centre for Sustainable Design (CfSD), and many more.
This strong collective environment capitalises on our resources, technological and clinical expertise, and is further bolstered by existing mechanisms to access regulatory and intellectual property (IP) expertise. Identifying joint IP that has been created and advising health boards on how that might be protected and developed, subject to any clauses in the collaboration agreement, is important from a commercial viewpoint and ensures clarity and transparency from all project stakeholders when it comes to implementation of AI in the NHS.
Scottish Health Innovations Ltd (SHIL) supports this work from a commercial point of view by advising health boards entering into collaborative agreements with industry partners to ensure board rights are protected. It also provides the necessary regulatory advice on navigating the evolving AI landscape.
The regulatory environment is fast changing and will continue to adapt to accommodate and support AI’s benefits. Time will tell whether the new UKCA mark can accelerate the pace of AI regulatory approval, but the MHRA consultation on the regulation of medical devices will hopefully result in a streamlined regulatory regime that supports Scotland’s ambitions to adopt AI at scale.
Data is vitally important. AI must be provided with quality data if it is to offer reliable results. The recent UK government consultation on healthcare data reforms (Data:a new direction) and the Goldacre review into the broader, safe use of healthcare data, are expected to further catalyse AI innovation.
It is vital to AI’s development as a new, potentially transformational technology that it is considered ‘trustworthy, ethical and inclusive’ – and these principles underpin Scotland’s AI strategy. Engagement with practitioners and patients is also crucial, including initiatives such as the Scottish Health Research Register and Biobank (SHARE) which involves citizens in validating research and technology.
Likewise, with an updated framework to be published in the autumn, the National Institute for Health and Care Excellence (NICE) will aim to ensure new technologies introduced into the health and care system are clinically effective and offer economic value by providing universal evidence standards.
AI undoubtedly lies at the heart of key NHS aims going forward – including encouraging new ways of working as part of a wider innovation ecosystem – and its implementation is timely indeed as the fight against cancer takes on fresh impetus at a vital time.
Fundamentally, Scotland is well positioned to support that AI evolution – and preventative opportunity – ultimatelyimproving patient experience, outcomes and of course saving lives.
