Some of the most cutting-edge artificial intelligence technologies in healthcare were showcased this week at an event in Glasgow.
‘Deep-learning’ algorithms which are being applied to analyse and predict cancers were demonstrated at the conference at the Queen Elizabeth University Hospital in Glasgow on Tuesday.
The Cancer Innovation Challenge Mesothelioma Project & AI in Healthcare event – hosted by The Data Lab – enabled some of Scotland’s most innovative tech companies, including Canon Medical, to demonstrate how new AI methods are helping clinicians to automate the process of identifying tumours.
Other ground-breaking applications of AI in health and care were also demonstrated – including for the improved treatment of patients with Chronic Obstructive Pulmonary Disorder (COPD) and those with suspected colon cancer.
Trade Minister Ivan McKee paid tribute to the work of specialists across Scotland’s ‘triple helix’ – a collaboration between industry, academia and the NHS – to develop pioneering new automated detection methods in a range of clinical settings.
He said: “We’ve put a lot of focus recently into our business research and development spend and that’s come on very well in the last two or three years, which really helps us to nail the whole innovation process down. The spend we’re doing on that is also coming through the industry academia links fund, which is where the money for the Cancer Innovation Challenge comes from and that really recognises the huge importance of the links between joining up what our great universities are doing and all the research work there and what business can bring to the party.
He added: “But it’s the impact it has on people’s lives which is the most significant factor and it’s great to see what’s happening. The Scottish Government is hugely behind this.”
Canon Medical demonstrated the work it has been doing with Dr Kevin Blyth, Consultant Respiratory Physician at NHS Greater Glasgow and Clyde, in developing a deep-learning algorithm to automate detection of mesothelioma, a type of cancer related to industrial exposure to asbestos.
Although the full results are yet to be published, the work carried out by the company’s data scientists has proved to be encouraging.
“I’ve been working in imaging for last 20 years and this has been the biggest step change in performance that I’ve seen during that time,” said Dr Keith Goatman, Principal Scientist, Canon Medical Research Europe. “And in cases like mesothelioma – where there is often disagreement between people about what’s mesothelioma and what isn’t – the consistency that this algorithm can bring can be very useful.”
RECIST (Response Evaluation Criteria in Solid Tumours) is a scoring system applied to CT scans to describe a patient’s response to cancer treatment, and is the gold standard measurement used in clinical trials.
However, RECIST is time-consuming and results often vary between reporting radiologists. A shortage of NHS radiologists also means RECIST is not routinely used in NHS care. Canon Medical’s aim is develop an automatic RECIST using Artificial Intelligence (AI), improving the quality and reducing the cost of cancer response assessment. The AI constructed will use deep convolutional neural networks (CNNs), after careful training using human inputs.
Dr David Lowe, Consultant in Emergency Medicine at Queen Elizabeth University Hospital, has also been working to devise AI methods which help predict when Chronic Obstructive Pulmonary Disorder (COPD) patients are likely to become breathless. Dr Lowe said that COPD is the second most common reason for people coming to hospital and that an AI project he has been working on could help better manage the condition in the community – leading to fewer hospital admissions and potentially saving the health service £1.2m.
He said: “Essentially we’re focusing on data, devices and decisions and really starting to think about how the NHS – the Queen Elizabeth (University Hospital] – can actually start to tell the story of how can engage in innovation to deliver products, services, solutions to enable us to deliver care better, more efficiently and improve patient outcomes.
He added: “In the space we operate in it’s really about how we assess whether that new technology, that new solution, is at a point for adoption. At what point should we jump and to be able to put it into care, demonstrate that it’s safe and that there’s a return on investment that allows us at board level to be able to say ‘let’s do this thing differently’.”
Dr Lowe said one of the difficulties of devising AI based training algorithms was that patients often have ‘stacked co-morbidities’ which all influence each other; however, he said the COPD project – which involved giving 400 patients Fitbits so they could monitor sleep patterns – he had worked on had produced ‘actionable insights’ and that the new Safe Haven Artificial Intelligence Platform (SHAIP) and the Industrial Centre for Artificial Intelligence Research in Digital Diagnostics will put in place a ‘structure’ whereby doctors can increasingly work with industry and academia to devise more data-driven service models.
“Getting a structure in place will enable us to do this and build that collaboration – so we’re able to say that this is what we’re capable of doing, and this is what we want to do,” he added.