In this blog Simon Elliman considers the possible application of artificial intelligence in detecting skin cancer, which is a growing problem in the UK and is going undiagnosed too frequently.
Melanoma is reported to be the fifth most common cancer in the UK with 13,500 new diagnoses every year, and the number of diagnoses is increasing.
Melanoma can be fatal and early detection is critical for maximising prospects of successful treatment. I have written in previous blogs about the seeming complacency by doctors in the UK in applying methods of detecting and treating melanoma, and my experience of cases where this has led to tragic outcomes for patients. I was therefore interested to read a recent article on ‘Artificial Intelligence’ being used to assist in the diagnosis of melanoma.
Melanoma is a form of skin cancer which generally carries a good prognosis if detected and treated early, and a very bad prognosis indeed if detected late. The most common symptom of melanoma is a change in the presentation of a mole.
The NHS have released an “ABCDE” guide to the examination of moles designed to highlight potentially worrying features warranting further investigation:
Any of these symptoms should be promptly and thoroughly investigated by a GP or other primary care practitioner, and further scans ordered where necessary.
Early diagnosis and treatment of melanoma usually leads to a successful outcome. The chances of survival dramatically reduce once the cancer has reached stage 3 and therefore any delays in treating patients in the early stages can have devastating effects.
However, as I reported last year there appears to be some complacency amongst GPs in the UK to act quickly and effectively.
Artificial Intelligence (AI) is the term used for when a computer seems to display cognitive behaviours belonging to humans, the most common being learning and problem solving.
Software designed by Google has been re-developed by scientists at Stanford University to teach Artificial Intelligence to identify the symptoms of melanoma.
The AI was shown 129,450 photographs of skin conditions and has been developed to ‘learn’ which were likely to be presentations of skin cancer. What is particularly interesting about the study is that when tested against certified skin cancer doctors, the AI came out on a par with them.
This development in technology could dramatically reduce the number of misdiagnoses and missed opportunities to detect and treat melanoma at the earliest opportunity.
Whilst it is not anticipated that the AI will replace the role of doctors and dermatologists in diagnosis, the two methods used in conjunction with each other could increase the likelihood of melanoma being detected in the early stages.
As a clinical negligence solicitor I see first hand where failing to diagnose melanoma promptly has led to devastating effects for my clients, which could have been avoided with early intervention. It is fascinating to see how technology that we can use in our daily life can be put to use in a clinical environment. I am very interested in developments which can increase early detection as prompt treatment would greatly improve the number of early diagnoses and therefore greatly improve the chances of survival of patients with melanoma, and will follow developments on this with interest.