A differential diagnostician
Over the past 10 years, AI has become a major ally of dermatology, providing new techniques that simplify the diagnosis and treatment of patients. There are many applications for which it adds tremendous value in dermatology: establishing precise differential diagnoses for common diseases, such as psoriasis, atopic dermatitis, or acne; eveloping personalized therapeutic protocols; and predicting medium- and long-term outcomes.
Furthermore, in onco-dermatology, AI has helped to automate the diagnosis of skin tumors by making it possible to differentiate between melanocytic and nonmelanocytic lesions. This distinction promotes early diagnosis and helps produce screening systems that are capable of prioritizing cases on the basis of their seriousness.
When asked whether any group has published any promising tools with good preliminary results, Dr. Conejo-Mir stated that his group has produced three articles that have been published in top-ranking journals. In these articles, “we explain our experience with artificial intelligence in Mohs surgery, in automated diagnosis, and for calculating the thickness of melanomas.” The eight-person research team, which comprises dermatologists and software engineers, has been working together in this area for the past 4 years.
Aesthetic dermatology
Unlike other specialists, dermatologists have 4-D vision when it comes to aesthetics, since they are also skin experts. AI plays a major role in aesthetic dermatology. It supports this specialty by providing a greater analytic capacity and by evaluating the procedure and technique to be used. “It’s going to help us think and make decisions. It has taken great strides in aesthetic dermatology, especially when it comes to techniques and products. There have been products like collagen, hyaluronic acid, then thread lifts ... Also, different techniques have been developed, like Botox, for example. Before, Botox was given following one method. Now, there are other methods,” explained Dr. Conejo-Mir.
He explained, “We have analyzed the facial image to detect wrinkles, spots, enlarged pores, et cetera, to see whether there are any lesions, and, depending on what the machine says you have, it provides you with a personalized treatment. It tells you the pattern of care that the patient should follow. It also tells you what you’re going to do, whether or not there is any problem, depending on the location and on what the person is like, et cetera. Then, for follow-up, you’re given an AI program that tells you if you’re doing well or not. Lastly, it gives you product recommendations.
“We are among the specialties that are going through the most change,” said Dr. Conejo-Mir.
An intrusive technology?
AI will be a tremendous help in decision-making, to the point where “in 4 or 5 years, it will become indispensable, just like the loupe in years past, and then the dermatoscope.” However, the machine will have to depend on human beings. “They won’t replace us, but they will become unavoidable assistants in our day-to-day medical practice.”
Questions have arisen regarding the potential dangers of these new technologies, like that of reducing the number of dermatologists within the population, and whether they might encourage intrusiveness. Dr. Conejo-Mir made no bones about it. “AI will never cut back the number of specialists. That is false. When AI supports us in teledermatology, even currently on our team, it spits out information, but the one making the decision is the practitioner, not the machine.”
AI is a tool but is not in itself something that treats patients. It is akin to the dermatoscope. Dermatologists use these tools every day, and they help arrive at diagnoses in difficult cases, but they are not a replacement for humans. “At least for the next 50 years, then we’ll see. In 2050 is when they say AI will surpass humans in its intelligence and reasoning capacity,” said Dr. Conejo-Mir.
Dr. Conejo-Mir has disclosed no relevant financial relationships.
This article was translated from the Medscape Spanish Edition. A version of this article appeared on Medscape.com.