Associate Feature: Acting on intelligence

“Could that growth on my skin be cancerous?” That is a question that prompts thousands of calls to GP surgeries every year.

Distinguishing between an innocent skin lesion and an unhealthy one is a skilled job, though, and not every GP has the training to make that judgment.

So more often than not, patients are referred to dermatology departments in hospitals for further investigation – which can mean long waits.

But what if it were possible, with the help of technology, to get a strong indication of diagnosis at the GP’s surgery in less than half an hour?

Dermatologists believe if such rapid indicative diagnosis were possible, it would improve outcomes for skin cancer patients, a key priority for NHS Scotland.

And so a new drive has been launched: “25 by 25”, to diagnose skin cancer in just 25 minutes by 2025, with the help of artificial intelligence (AI).

It’s being pursued by a consortium led by NHS Scotland and bringing together industry figures and academics.

The group are determined that within four years, by integrating AI into the patient care pathway with the expertise of medical staff, it will be possible to provide that indicative diagnosis in less than half an hour across Scotland.

Dr Colin Morton, consultant dermatologist at NHS Forth Valley and clinical lead for the consortium, says: “Emerging evidence points towards the potential for AI to help us reach quicker diagnosis of skin cancer.   

“Many patients should also be able to receive quick reassurance where a new growth can be shown to be innocent.

“We still have a way to go to develop and thoroughly test a fit-for-purpose AI platform, but we hope that, if supported by national funding, the citizens of Scotland can be amongst the first to benefit from using this technology.”

Over half of the work done by specialist dermatology services is in the diagnosis and management of suspected skin cancer, but prior to the pandemic dermatology teams were already struggling to meet the 62-day national target for cancer diagnosis and treatment.

The pandemic has created added pressure as medical staff struggle to deal with a backlog of cases.

Having specialists make diagnoses using high-quality images of skin lesions (instead of having to see the patient in person) is the first important step in speeding up cancer diagnosis while the NHS awaits AI solutions. If, when patients are referred to dermatology departments with suspected skin cancer, high-quality pictures of the growths in question are attached, it can allow the teams to identify which patients on the waiting list are most likely to have skin cancer so they can see a specialist more quickly.

In certain parts of Scotland, models of care have been developed which make routine use of images.

In Tayside, 90 per cent of referrals to dermatology departments by GPs already have an image attached for more efficient diagnosis. Professor Colin Fleming, database and research lead for the consortium and professor of dermatology at NHS Tayside, says that around 6,000 patients referred every year – one third of the total – can be provided with a diagnosis and treatment plan within just two days.

The consortium would like this to be extended to all of Scotland.

But images of skin lesions sent to GPs by patients can be of highly variable quality and so work is already underway to set up novel community and local imaging centres (CLICs) across the country to help ensure high quality images can be taken at primary care level, while avoiding passing on additional work onto GP surgeries.

The AI technology would build on this widespread use of images and provide a step-change in improving skin cancer diagnosis and care across Scotland.

Initially, the consortium expect the technology would be used to help recognise vital signs of cancer change from pictures. One likely big use of it would be to provide reassurance for patients and reduce long waits for specialist advice. But importantly, the AI would provide an indicative, or likely, diagnosis to improve efficiency: patients would still be assessed by specialist teams and progress to surgery where required.

In time, the team hope it might be possible to enhance the power of the AI platform to predict how a patient’s condition could progress.

They hope to design the technology not only to analyse skin images but to take into account information from the patient’s history.

All in all, the consortium believe it would be a “leap forward in healthcare”.

They have identified three priority settings for the use of the technology. Helping diagnosis of skin cancer is one. The second is in helping patients self-monitor their symptoms. For instance, an app using AI could help patients take high quality images on a regular basis to monitor their own moles, helping them spot changes that may need an expert review. The third priority would be in making treatment decisions.

There is a pressing need to speed up the diagnosis of skin cancer and improve treatment and care. At least 100,000 cases are diagnosed across the UK every year and rates are rising.
Melanoma kills more young adults than any other cancer in Scotland.

In Scotland, around 12,000 people are diagnosed every year with non-melanoma skin cancer, of which around 3,000 are squamous cell carcinomas.

On top of that, around 1,200 malignant melanomas are diagnosed each year.

If melanoma is caught early, the five-year survival rate is around 95 per cent. If it is not caught early, that survival rate can drop to around 50 per cent.

The Covid-19 pandemic has created added pressures.

There was a 24 per cent decrease in the number of cancers seen between April and September 2021, and during that period there were 34 per cent fewer melanoma cases being seen than would normally be expected.

Dermatologists are worried that patients are presenting later, when their cancer is more advanced, which underlines the importance of specialists getting access to images of growths associated with patients on the waiting list, so those who need rapid assistance can get it.
The “25 by 25” target echoes the Scottish Government’s desire for NHS Scotland to emerge from the Covid-19 pandemic stronger than before, using technology where possible to advance the diagnosis, treatment and care of patients.

NHS Scotland’s Remobilise, Recover, Re-design framework aims to find ways of building on some of the new ways of delivering care borne of necessity during the pandemic.

The NHS recovery plan published in August by new health secretary Humza Yousaf stressed the Scottish Government’s support for “innovation in and redesign of services to ensure that more patients receive person-centred care in the right place, at the right time”, while committing to accelerating the adoption of new technologies that support improved access to healthcare – including AI.

The consortium goals are ambitious but they believe that with investment they can quickly improve the capture of images, by setting up CLICs, during 2022, and then working with universities and industry to develop and test the AI so that it can be introduced into clinical practice by 2025, noting “the clock is ticking”.

Whilst the consortium are focused on supporting the rapid, accurate diagnosis of skin cancer, they also hope their work will help to identify and explore the potential for AI and innovative solutions in other high demand areas.

As for Dr Morton, he says: “Appropriately regulated and governed uses of AI could enhance safe clinical practice and improve patient outcomes. AI has the potential to be part of a joined-up integrated pathway of care, still involving the healthcare professionals we are familiar with, but helped to become more efficient and cope better with soaring demand.”

The consortium believes AI will be a “game-changer” in how healthcare is delivered in a number of specialities: “Already AI is significantly used in other parts of everyday life, for instance when we use computer search engines and in vehicle technology.

“Its potential to help us deliver healthcare is huge, but needs to be directed to where it can be of greatest benefit.”

They hope that its first benefit will be in improving outcomes for the growing number of people in Scotland with skin cancer. 

This special feature was written in association with Scottish Health
Innovations Ltd (SHIL), West of Scotland Innovation Hub and the Scottish Dermatology AI Innovation Consortium. The consortium is led by the Innovation Team in NHS Forth Valley and brings together the Centre for Sustainable Delivery, the Scottish Health Industry Partnership and the Cancer Access Team. It is supported by the Chief Scientist Office of the Scottish Government and includes a range of industry and academic partners including iCAIRD, The Data Lab, Digital Health and Care Innovation Centre.