Artificial intelligence (AI) is no longer an alien concept. It’s shaking up traditional perceptions and saving precious time across different sectors on a global scale – and there’s no doubting its vast potential for healthcare innovation in Scotland too when it comes to improving patient experiences, outcomes, and of course saving lives.
In fact, it’s already starting to positively impact everything from diagnostics and preventative medicine to the development of devices and pharmaceuticals, while encompassing ethical and inclusive goals.
AI represents nothing less than a huge opportunity to radically change health and social care – and quite simply we cannot continue the renewal and transformation process with human resource alone.
Technical and clinical expertise is now combining to tackle major health challenges, and Scotland’s ever more collaborative ecosystem offers a rich environment in which to work with industry to develop solutions that improve the quality, efficiency, and sustainability of healthcare.
I’m absolutely convinced that there’s no better place in the world to deliver a healthcare renaissance with AI, than Scotland.
In many respects, that admittedly slow-emerging renaissance was catalysed by the Industrial Centre for AI Research in Digital Diagnostics (iCAIRD) programme. Our fundamental aim was to draw together and coordinate technical, research, healthcare, and industrial capability from across Scotland to build solutions that solve real world, high priority healthcare issues with AI. We wanted to show what could be achieved, so that others would follow – and we did.
Programmes like iCAIRD demonstrate that with the right data at our fingertips, and the NHS working hand-in-glove with innovators to robustly validate technology, Scotland can invent and implement meaningful AI solutions.
It began with 10 partners, 10 projects, and £10m and grew to be the largest healthcare AI research programme in the UK with 40 partners, 50 projects, and £25m in public and private funding. It was a real demonstration of enthusiasm, collaboration, and support that saw us win six national awards.
iCAIRD helped securely and safely deliver an unprecedented 100+ million medical images and associated patient data for research aims; built platforms for AI research and evaluation; created one of the largest digital pathology labs in the world; spearheaded three market-ready products; scaled four SMEs and established a national AI evaluation team.
We are proud of the volume and breadth of imaging that has been tapped into, particularly for radiology research purposes – including over a million chest x-rays, over 100,000 mammograms, over 50,000 head CTs, and thousands of MRIs. This data was already there but could not be accessed until iCAIRD.
Our pathology research and validation meanwhile went from zero to 20 projects in a year-and-a-half, and the digital pathology archive we created at NHS Greater Glasgow and Clyde now has over two million histology slides. In combination with the data already accessible via Scotland’s Safe Haven and Biorepository networks, it makes Scotland one of the global go-to locations for multi-omic data. That kind of data underpins the discovery of new medicines and treatments for a wide variety of diseases.
Whilst iCAIRD is now gone, its legacy remains. Spin-off projects are tantalisingly close to realising our goal of using AI to improve the quality and accuracy of diagnosis in lung and breast cancer. The regional healthcare innovation hubs in North, West and South East Scotland have more AI projects than ever before. Furthermore, health boards are looking beyond radiology towards using AI in an ever-increasing range of clinical and non-clinical areas.
Scotland has a national imaging research capability called the Scottish Medical Imaging service, which is run by the electronic Data Research and Innovation Service (eDRIS), part of Public Health Scotland, and works with Research Data Scotland.
However, AI adoption is moving slowly because we need to gather more real-world evidence that AI products are clinically effective and deliver value within complex and often inconsistent clinical pathways.
I believe that with the right infrastructure, strategy, and support from government and the healthcare sector, we can achieve a tipping point over the next two or three years and move from research to widescale adoption.
With virtually every health board in Scotland using AI in some form, and industry and academia keen to collaborate and pioneer, significant breakthroughs are happening all the time.
One leading example that has made headlines is AI software known as Mia which is being trialled in detecting signs of breast cancer which can be extremely hard to identify with around 20 per cent of women with breast cancer tumours missed by mammogram screening.
Mia helped doctors find an additional 12 per cent more cancers than in routine practice. Put in a wider context, Mia – a collaboration between Kheiron Medical Technologies, NHS Grampian, the University of Aberdeen, and Microsoft – could therefore lead to better outcomes for thousands of women across the UK if deployed across the entire NHS.
Dr Gerald Lip, who led the prospective trial at NHS Grampian, further noted that the AI had modelled a workload reduction of up to 30 per cent thanks to an expected decrease in women being recalled unnecessarily for further assessment.
Around 48,000 people living in Scotland have been diagnosed with heart failure by their GP, with thousands more thought to be living with the condition unknown and early diagnosis considered vital to reducing the risk of hospitalisation. In combination with more traditional improvements to the heart failure pathway, The Optimised Pathway for Early Identification of Heart Failure in the Community (OPERA) was able to reduce waiting times from 12 months to two weeks.
Live clinical evaluations of chest x-ray AI are currently underway in both NHS Grampian and NHS Greater Glasgow and Clyde. AI is used to spot possible signs of lung cancer, Scotland’s most common cancer, in a patient’s x-ray. These patients are then prioritised for definitive diagnosis and treatment.
Early evidence shows that we could reduce the time between getting chest x-rays and being treated for lung cancer by around nine days and increase the diagnosis of early-stage lung cancer by 11%. That could mean 600 more people per year in Scotland being diagnosed early enough that their lung cancer can be treated.
Such is the promise of this technology that the case for national adoption is being examined by the Accelerated National Innovation Adoption (ANIA) collaborative. Given that 65% of lung cancer cases are diagnosed at stages 3 and 4, where the chances of survival are very small, anything we can do to improve those odds must surely be welcomed.
There are many more examples of live evaluations using AI across the Scottish healthcare system, including bone fracture detection, accelerated MRI image acquisition, discharge management, hospital readmission management, accelerated detection of head trauma, mental health, and the management of long-term conditions such as Chronic Obstructive Pulmonary Disease (COPD).
Regarding national adoption, it will no doubt surprise many to know that there is currently no single AI tool or technology used ubiquitously across Scotland’s territorial health boards. AI is being used piecemeal in areas such as paediatric growth monitoring, dynamic radiotherapy planning, paediatric cardiology, and radiology. However, until now there hasn’t been a nationally coordinated adoption project.
Stroke remains one of Scotland’s biggest killers and leading causes of disability. That’s why it’s great news that Scotland is procuring a national stroke AI solution that might halve the door-in-door-out time for stroke patients and triple the number who will recover with no or slight disability.
Of course, in many respects, the application of AI to solve health and social care problems remains a young, rapidly developing field. It is important therefore that we invest in the innovation lifecycle from end to end, including research and development, skills development, business growth and support, and robust evaluation within the clinical environment.
It is not enough for innovation in this area to be novel – it must be viable, evidenced, safe, fair, and properly embedded into clinical practice. We have to clearly describe the most critical problems and focus our attention on fixing a small number using a combination of AI, automation, digital, workforce and service transformation. Our approach should be consistent across health boards, coordinated nationally and targeted at whole clinical pathways, not just individual services or specialities.
To maximise the true potential of AI healthcare, policy, finance, and industry leaders need to come together to provide their expertise and insights on how we remove barriers, build trust and scale quickly using innovative procurement models to tackle the thorny issue of not having the money for the traditional NHS invest-to-save approach.
We have to make this happen quickly. It is already taking too long – lives are at stake.
Where Scotland is well placed to excel in AI is through top-quality academic and commercial research capability as well as trusted, enduring relationships between NHS, industry, and universities across a talent-filled ecosystem with opportunities for Scottish spin outs and start-ups to expand and dominate the international scene.
AI is set to change the way we all live, and that includes our healthcare system. I have no doubt that Scotland’s innovators can help to save lives by harnessing its incredible potential.
JD Blackwood will be a guest speaker at InnoScot Health’s webinar on 29 May, entitled AI in Healthcare: IP and commercialisation