AI and the Future of Cosmetic Dermatology: Experts Weigh in

While artificial intelligence (AI) is helping clinicians in medical dermatology improve diagnostic accuracy, streamline workflows, and develop patient treatment plans, its wide adoption for use in cosmetic dermatology is still in the early stages.

According to Jeffrey Dover, MD, co-director of SkinCare Physicians in Chestnut Hill, Massachusetts, cosmetic dermatologists are primarily using AI for patient management in the same way as their counterparts in medical dermatology. Dover and his team use AI-powered software to “select the right appointment duration for the right procedure or the right kind of visit,” he said in an interview. And on a basic level, “we also use it to communicate with patients about reminders for their appointments, which decreases the number of same-day cancellations and no-shows.”

Among the cosmetic dermatology applications currently under study are image recognition techniques using deep learning (DL) algorithms to measure the water content in a patient’s skin, which provides a key parameter for assessing skin quality both before and after treatment. Also, DL algorithms, combined with computer vision technology, have been used to identify fine lines in patients, helping to determine the most effective treatment options.

In the realm of skin analysis, AI-assisted imaging technologies like LifeViz Micro and the VISIA Skin Analysis System can be used to evaluate various skin features such as wrinkles, pores, pigmentation, skin texture, and ultraviolet damage, which helps professionals recommend personalized treatments or skincare regimens. Both platforms also generate data to help measure outcomes of recommended treatments.

However, the definition of “success” can be elusive, Dover said. He recalled a conversation with a plastic surgeon who used AI software to assess the degree of improvement after performing a facelift on a female patient. “He took all these measurements, and the patient showed a 94% improvement as assessed with the AI application in attributes like sagging jowls, forehead drop, brow ptosis, and cheek flatness,” Dover explained. When the surgeon shared these results with the patient, she replied, “That’s terrific, but I don’t like the way it looks.”

This example highlights the challenge of applying quantitative analysis to beauty. “As the adage goes: ‘Beauty is in the eye of the beholder,’” Dover said. “What patients expect and what providers aim for must align, and while AI software can be a useful tool, it can also get in the way.”

Finding a ‘Seat at the Table’

Fatima N. Mirza, MD, chief dermatology resident at Warren Alpert Medical School of Brown University in Providence, Rhode Island, acknowledged that while there are “growing pains” associated with the introduction of new AI technologies in dermatology, clinicians can influence not only how these technologies are ultimately integrated but also the quality of the data used to train the models. In a 2024 study published in JAMA Surgery, Mirza and her colleagues used three publicly available AI text-to-image generators — a type of AI system that creates images based on textual descriptions provided by the user — to gauge the demographic accuracy of surgeon representation compared with real-world attending surgeons and trainees.

The researchers found that all three models underestimated trainee representation and revealed certain biases. For example, two of the models depicted more than 98% of surgeons as White and men. “When we asked the three models to produce the photo face of a Mohs surgeon, two did not produce an image of a single female physician,” added Mirza, who has been given early and exclusive access to some emerging technologies from OpenAI, the developers of ChatGPT. “However, some of the folks who make these models read our study and were able to adjust the weights of their models to address those biases. I think that’s why it’s important for clinicians and patients to have a seat at the table, to be able to adjust how these models continue to evolve over time.”

Meanwhile, a company called BestLook.ai has developed software that analyzes facial features such as symmetry, proportions, and overall “attractiveness” based on AI algorithms that have been trained on large datasets of facial features. “It suggests changes to your face that I could potentially make during your visit, such as increasing cheek fullness, making your lips fuller or smaller, narrowing the bridge of your nose, or reshaping the ala,” explained Dover, who is consulting with the company to explore its applications in aesthetic medicine. “You can use a split screen to toggle between the before-and-after images. It’s fascinating.”

Privacy Concerns About Images

While the potential of AI to assist in the delivery of cosmetic dermatologic care is clear, it also raises questions and concerns. Dover noted that some people find it “creepy” for AI to determine their ideal look, especially when it comes to aesthetic choices. “In current practice, they’re relying on the doctor or nurse practitioner to help them make the decision,” he said. “But what if their ideas are off base from what the patient wants?”

This concern over AI’s role is reflected in the reluctance of some people to use the software. One of Dover’s research fellows, for instance, found that several people refused to let her take pictures and upload them to BestLook.ai, fearing their images might end up in the wrong hands. “They didn’t trust the security of the software,” Dover explained. “Interestingly, some of those individuals were prominent figures in the dermatology industry who also refused to participate,” he said. “AI systems cannot operate autonomously; they must be linked to clinical oversight and have secure, protected walls that are impenetrable. Every day, we hear about companies experiencing data breaches. When it comes to your credit cards, that’s one thing. But when it’s your face, that’s a whole different story, isn’t it?”

Another product, a smartphone app for practices known as CherryPics, generates three-dimensional images of the face or the body of patients that can be used to track changes following aesthetic procedures over time. According to Arisa E. Ortiz, MD, director of Laser and Cosmetic Dermatology at the University of California, San Diego, and current president of the American Society for Laser Medicine and Surgery, this app “is good for volumetric analysis for, say, pre- and post-filler treatments, but it can be more difficult to assess pore size with.” In her practice, she continued, such AI-driven technologies “are more of a research tool for me because trying to [use them] for every patient when you’re seeing dozens of patients in a day is just not practical. But if a patient wanted to assess these things, it’s nice to have them available.”

Ortiz expressed concern about AI programs that use augmented reality to generate an image of what a patient might look like after a procedure. “They might give patients a false sense of what we can achieve as cosmetic dermatologists,” she told this news organization. “It can be helpful for the patient to visualize what the outcome may look like, but I think it sets unrealistic expectations — like photos that are corrected with editing tools such as filters and Facetune. I have concerns with apps like that.”

While Ortiz raises concerns about the potential drawbacks of AI in shaping patient expectations, she also sees its value in improving the efficiency of consultations and providing patients with tailored, informative experiences. This perspective inspired her to work with a software engineer to develop Your Previsit, an app that patients can download to their smartphones, featuring AI-generated videos specifically tailored to the services of cosmetic dermatology practices.

The inspiration came from Ortiz’s own frustration at the time-consuming task of explaining procedures over and over during patient consultations. “I have a script in my head that I tell my patients about each procedure. I thought, ‘I wish there was a video of me saying this, that the patients could just watch,’” she said. “When patients come in for consultation, they are so overwhelmed by the amount of information that I have to really limit what I talk to them about to two or three procedures.”

The app uses content curated by Ortiz to create short educational videos about procedures that patients can watch before coming into the office. “For example, if they want to learn about wrinkles on the face, the app will generate a list of 3- to 4-minute procedural videos offered by the practice that they can watch on their own time,” she explained. “The videos cover the procedure’s purpose, how it works, expected downtime, pain levels, how long it takes to see results, and more. This way, by the time they come in for their visit, patients feel more informed, empowered, and ready to ask more focused questions.” The app, which launched in December 2024, allows providers to edit content scripts, ensuring personalized AI-generated videos tailored to their practices. There’s also an option to add a custom avatar — a “digital twin” that uses AI to mimic the provider’s image and voice — with no need for costly and time-consuming filming. “That has been a barrier in the past; few clinicians have the time to go to a recording studio, create the script, and actually film the video,” Ortiz said. “The app eliminates this hurdle by using AI to generate both the script and the videos.”

Improving Patient Communication

As another example of how AI is being used to improve communication between clinicians and patients, Mirza and colleagues utilized ChatGPT-4 to create an AI-generated surgical consent form for Brown University Health, Rhode Island’s largest healthcare system. The original consent form was written at a college sophomore level and was then translated by ChatGPT into language more accessible to the average American reading level, which is around the sixth to eighth grade, she explained. In the study, Mirza and her co-authors noted that their work demonstrates AI’s potential to enhance communication, reaching vulnerable patients who are often left behind by the overly complex and verbose language of traditional medical forms.

While clinicians explore how to most effectively integrate AI into their practices, Mirza remains optimistic about its potential to improve both patient communication and the delivery of optimal dermatologic care. “I think it’s an exciting time to be alive, to be at the intersection of medicine and AI,” she said. “As these models improve, they might even be able to provide us with physician-specific outcomes of a particular cosmetic procedure. Although the technology may not be fully developed yet, this is an exciting opportunity for us to start the conversation and, as clinicians and a community of physicians, consider how we can contribute to improving these advancements over time.”

Ortiz holds financial interest in Your Previsit. Dover is a consultant for BestLook.ai. Mirza reported having no relevant disclosures.