Consultant Dermatologist and Dermatological Surgeon, Dr. Magnus Lynch, on the application of artificial intelligence methods within dermatology and how it can empower patients…


Dr. Magnus Lynch is a London-based Consultant Dermatologist, Mohs Surgeon and Laser Surgeon. He is highly experienced in all aspects of adult dermatology with particular expertise in skin surgery and the diagnosis and treatment of skin cancer.

Dr. Lynch was a scholar at Christ’s College, Cambridge and graduated in medicine in 2001. He subsequently passed membership examinations for both the Royal College of Physicians and The Royal College of Surgeons and then trained under leading dermatologists at the world-renowned St John’s Institute of Dermatology (Guy’s and St Thomas’ Hospitals), gaining entry to the GMC specialist register (the UK equivalent of board certification in Dermatology) in 2017. After completion of training in Dermatology, he undertook a highly competitive fellowship in Mohs surgery and advanced dermatological surgery at St John’s where he is now a Consultant Dermatologist and Mohs Surgeon.

Boasting a strong academic background, he undertook a PhD at the Institute of Molecular Medicine, University of Oxford, was awarded a prestigious Academic Clinical Lectureship at St John’s Institute of Dermatology and has performed postdoctoral research at King’s College London. He is currently a Senior Lecturer at King’s College London where he leads a research team investigating cellular biology of the skin and the role of AI in dermatology.


What initially sparked your interest in medicine and subsequently, AI in medicine?

I have always been interested in science and how the body works. But I then started to question why disease occurs and what we can do about it, and that was what sparked my interest in medicine. After completing my initial training in internal medicine, I went on to do a PhD in epigenetics. This was at a time when high throughput sequencing technology was at its infancy and bioinformatics and coding formed a large part of the project requiring me to learn new skills. I continued to develop these skills over time, including during my postdoctoral research, and as I continued with my specialist training in medicine and dermatology, I started thinking about applying the methods I have learnt, including machine learning and deep learning and how we could apply these to questions we encounter in the clinic to help us better stratify and treat patients.

Did you ever consider another career path?

No, medicine has always been something I was keen to pursue. Of course, there have been challenging times, having to balance multiple demands from research to patient care. One exciting thing about medicine is that you are always learning new things and encountering new challenges. In the early phases of your career the challenge is to master the existing body of knowledge. However, as you progress, there is the potential to contribute new knowledge by introducing new ways of doing things and that’s truly exciting.

What are you most excited about the future of AI in dermatology?

To date, AI in dermatology has largely been applied to skin cancer, particularly magnified (dermoscopic) images of patient’s skin. There has been a lot less work in applying these methods to clinical images of patients and pathology images. Not only for diagnosis but also for disease stratification and prediction of response to treatment.

There is still an enormous number of clinical images and pathology slides which are produced on a daily basis; however these are siloed in respective clinical departments. A coordinated national strategy has the potential to allow – with patient’s consent – access to this data and the development of new AI-powered approaches to diagnosis, disease stratification and monitoring of treatment response.

Aside from data, are dermatologists receptive to AI in general?

Most dermatologists that I work with are very receptive to AI and indeed are enthusiastic about the potential. Many people do not realize that making a diagnosis is only a small part of the work that dermatologists do. Other aspects include discussing the diagnosis with patients, explaining when there is uncertainty in the diagnosis, discussing the pros and cons of different treatment options and working with patients to agree a treatment plan. For this reason many dermatologists do not see AI as a threat but rather as a powerful tool that has the potential to empower patients and to permit more efficient and accurate methods of disease stratification and treatment monitoring.

Who has been the biggest influence on your career?

Two people really impacted my academic work – Professor Douglas Higgs, who was my PhD supervisor and Professor Fiona Watt who supervised my postdoctoral work. In terms of clinical work, I couldn’t name a specific individual since so many people have helped me over the years.

What do you consider your biggest achievement and failure?

My biggest achievement is to combine my clinical work with my academic work. This has been challenging since there are multiple and conflicting pressures from both sides. Nevertheless, being able to help patients both directly and to contribute to the development of new knowledge that may permit novel treatments for patients in the future is incredibly satisfying. In terms of failure, perhaps I could have recognized the importance of AI within clinical medicine earlier.

You mention resources. What kind of support would you have liked to receive to help you get involved in AI earlier?

I have a lab and many ideas for research, however in common with many scientists I do not have sufficient resources to pursue all of the ideas that interest me! I would definitely benefit from having more individuals within my lab with skills that enable them to answer questions at the intersection of AI, cell biology and clinical medicine.

How challenging is it for someone with a medical background to step in and start learning and developing something AI related?

It is challenging and usually will require a considerable period of time dedicated to building the skills required – this could be during medical school or during a period of research such as a PhD. That said, there are many resources online for learning both coding and AI development and most motivated individuals with sufficient time and effort can master these concepts. Of course, not all medical professionals need to be experts in machine learning and for many clinicians the optimal course of action will be to work in collaboration with an AI researcher.

What advice would you give to someone starting their career in medical AI?

If you are interested in AI, it is ideal if you can start building skills in coding, statistics and data analysis whilst in medical school. This could be either via formal taught courses or self-directed learning online. Even if your goal is not to become an expert in coding or AI these skills are becoming more and more essential to any form of biomedical research or analysis of large-scale clinical datasets.

What’s the best piece of advice you’ve ever received?

To identify the problem that you want to solve, work out a plan and then stick with it until completion. Sometimes it’s tempting to jump around a lot when a more interesting project comes along but this can be counterproductive.

I guess it’s similar in life too. One should aim to finish the thing you have started, regardless of whether it will change the world or not. You build confidence from achieving small goals, projects or research tasks and that confidence allows you to take on bigger challenges.

What’s your greatest fear about the development of AI?

That patients are not sufficiently involved in the development of medical AI. It’s so important that patients are actively involved in discussions of potential applications and that informed consent is obtained when patient data is used for the training of AI models. If we are not mindful of these factors, we may lose the confidence of patients.

How do you think we can do better?

Within dermatology in the UK, I think there is a need for a national system for the collection of data from patients that consent to their clinical images and other data being used for the training of AI models. We need to have dialogues with patients and be open to them about what we are going to do with their data and how it will be used. Establishing this will involve a considerable amount of work requiring ethical approval and engagement with multiple stakeholders.

Do you think we should also inform patients that AI is part of their diagnosis or treatment process?

I think this is essential. As mentioned, we should be open to patients about what we are doing, why we are doing it and what are the benefits. I believe in certain cases – for example, quantifying disease severity – AI can potentially perform better than a human.

Do you think AI will aggravate health disparities or improve them?

I think in the long-term AI has the potential to improve health disparities by removing barriers to access in healthcare. However, we need to be very careful about how technology is deployed and validated.

Dr. Magnus Lynch will be speaking at AIMed’s virtual multi-track CME-accredited event, ‘Imaging’ on 29th and 30th June.

View the full, exciting two day agenda and book here