An NHS Scotland-led consortium will be launched today with the ambition that, using artificial intelligence, by 2025 it will be possible to diagnose skin cancer in less than 25 minutes.

Skin cancer is the most common cancer in Scotland, and melanoma – which represents around 10% of skin cancers – is the deadliest cancer among young adults.

Swift diagnosis saves lives. While 95% of people whose melanoma cancer is spotted early will still be alive five years later, this drops to half if there is a delay in treatment being started.

One of the people behind the AI Skin Cancer Consortium, Lead Consultant Dermatologist at NHS Forth Valley, Dr Colin Morton, has been telling healthandcare.scot of the opportunity that Scotland has to embed world-leading artificial intelligence into patients’ journeys, to transform skin cancer diagnosis and offer support – and reassurance – to both patients and healthcare professionals.

 

The diagnosis and care of skin cancer patients represents half of the workload of Scotland’s dermatology specialists and, with the incidence of skin cancer increasing at the rate of 5% per year, the demand for hospital dermatology appointments is outstripping capacity.

“We have for many years as a specialist team been doing the mouse wheel thing of running faster and faster to try to meet demand,” says Colin Morton.

One in five GP consultations are around skin complaints and, as a generalists, doctors and nurses working in the community prefer a specialist to look at a skin lesion to be sure it is not cancerous.

The Scottish government expects the time from referral to treatment to be less than 62 days – a target that Dr Morton says was becoming harder to meet even before the pandemic.

“We have reached the point where we simply do not have sufficient workforce to satisfying the demand that is coming in. We need to look at a different way and this is where the great opportunity from artificial intelligence comes in.

“What we want to do is to use artificial intelligence to help us in our diagnosis, our care and management of suspected skin cancer here in Scotland.

“It was only in 2017 when an article in the publication Nature suggested that digital solutions had an ability to spot cancers with a similar level of accuracy to an established dermatologist. There have been quite a number of publications since then, but these are often in quite contrived situations – they were done in a research setting. What we are really keen to do in Scotland is to bring this into real-world use.

“Already you can download apps on your smart devices but, as far as I am aware, none is approved for diagnostic purposes.

“If we are being serious about taking the use of artificial intelligence forward in our care of patients, we really need to move up a gear and have technology certified for use for diagnostic purposes.”

A national database

Colin Morton says the AI Skin Cancer Consortium project relies on drawing together images from across Scotland’s NHS into a national registry to “teach” artificial intelligence systems to “learn” what cancerous lesions look like.

One of the priorities is to gather images of both cancerous and benign lesions into a single accessible database.

“There is a well-worn phrase in the current conversation of ‘the value of data’ and we have a lot of it across NHS Scotland.

“Dermatology has always been a specialty where images are important and for many years we’ve been taking and receiving images of suspected cancer lesions in patients as a way of us helping to triage referrals – where a professional, using human time and expertise, looks at that image and decides whether the patient should be coming urgently to hospital or not.

“Scotland is sitting on a large amount of data both in dermatology departments and in patients’ own notes. The key however is that every image needs to have a diagnosis linked to it.

“We can anonymise the images, tagged with the correct diagnosis, and start joining everything together within the NHS’s secure systems. Having that fuel will really promote the development of these machine learning algorithms. Being selfish, working in Scotland, we need to be able to use Scottish images to train the system to recognise cancerous lesions on local skin types.”

Dr Morton says by coming together as consortium led by the NHS and government’s Chief Scientist Office, bringing in the expertise of universities and companies, with the hope of greater involvement of the third sector and patients, he is confident that the AI Skin Cancer Consortium will be able to deliver solutions that can be used at several points along a patient’s journey from spotting an unusual skin growth, through diagnosis, treatment and getting the all-clear.

“The leadership from NHS Scotland is so important because what we are looking for is not some sort of stand-alone digital solution that ignores all the things that humans can bring to the health care pathway: offering reassurance and helping patients to understand what it’s all about. We are looking to combine artificial intelligence with people to give us substantially improved capacity, shortening the recognition to diagnosis journey, creating a fit-for-purpose modern service that can diagnose skin cancer quickly.”

25 by 25

What then might skin cancer services look like in 2025 if machine learning has allowed artificial intelligence to provide a reliable diagnosis of a potential skin cancer within 25 minutes?

The programme will focus on three touch points where the team believe artificial intelligence can be transformative: for the patient at home, the GP and the dermatology specialist.

For the person at home, spotting a mark or rash on their skin, Colin Morton says people will be able to use a tested and certified app:

“You can envisage that you would use the camera on your smart device pointing at the lesion and an app downloaded from NHS Scotland to give you some indication of what you’re looking at. That’s more about some early reassurance or encouraging a patient to seek advice.

“And, if the app can also collect information about your age, whether you lived in Australia for five years and other risk factors, then that will help when you go to your GP.

“The second touch point is probably the most important in terms of delivering the service that we need in Scotland skin cancer care, and that is developing a diagnostic aid for GPs and allied nurse professionals in primary care so that, if they are looking at a skin growth that they are not quite sure about, they can use this technology and it will indicate a probability score, with a good degree of accuracy, that what they’re looking at is something that needs further investigation or, alternatively, allows them to reassure a patient there and then that it is an entirely benign growth.

“And then, finally, we recognise that we can also utilise artificial intelligence within the specialist community, with the information from an image of a skin lesion allowing us to get it right and to direct the patient to the correct choice of treatment first time.”

The consortium, to be chaired by Janie McCusker, Chair of NHS Forth Valley, will bring together NHS boards, Scotland’s innovation hubs, universities and companies to support the development of a national skin images registry, run a program of innovation challenge competitions and coordinate research and development projects leading to the testing and trialling of new AI tools in Scottish healthcare.

 

Sign up to our bulletin sent twice a week so you don’t miss out on the latest health and social care news.