Physician Spotlight: Steven R. Feldman, MD, PhD

Will artificial or augmented intelligence (AI) technology ever replace physicians? Steven R. Feldman, MD, PhD, Professor of Dermatology, Pathology and Public Health Sciences at Wake Forest School of Medicine in Winston-Salem, NC, spoke with Practical Dermatology® magazine about the role for smartphones, apps, and machine learning.

Tell us about the study you just conducted on Perfect Corp.’s AI Skin technology.

Steven R. Feldman, MD: We compared Perfect Corp.’s AI Skin technology to both physician assessments and a high-end, standard imaging device and found that AI Skin had good correlation with assessments done by a physician and the clinical imaging system with dedicated hardware and camera. AI Skin is also really fast, making it fun to use. Perfect Corp.’s AI Skin software was able to deliver results much more quickly than the clinical imaging system, which took about 3 minutes. The new research serves as the first step toward demonstrating that AI Skin is a reliable way to assess skin outcomes.

How does this software work?

Dr. Feldman: It is fabulous because your iPhone or iPad is all that is needed to do the work. It takes 8 seconds to capture an image and get a score. You don’t need to invest in a fancy system. AI Skin allows us to quantify facial characteristics and show patients changes over time in a quantitative way. We can quantify success with treatments and quantitatively document improvements, thus demonstrating the value and quality of our services. It’s hard to improve what you don’t measure.

The AI face analyzer provides users with real-time online analysis of their skin health in 14 parameters including moisture, spots, wrinkles, dark circles, oiliness, texture, acne, redness, and pores. Scores based on the severity of these conditions can be combined to calculate the user’s estimated skin age. The technology also differentiates skin conditions at different stages. It is designed primarily for cosmetic assessment of things like redness, wrinkles, lustrousness, and pore size.

How does it come up with this score?

Dr. Feldman: It’s more of a black box than an algorithm. The technology learns the same way we do. You give it a lot of examples, and that teaches it how to think.

So, will AI replace dermatologists?

Dr. Feldman: I used to think we would need a dermatologist to make a diagnosis. But an iPhone can see better than a dermatologist’s eye, seeing wavelengths we cannot see, and has greater processing power. Add that to artificial intelligence learning, and technology should be able to make diagnoses and offer treatment options at least as well as a human can. A computer can beat Magnus Carlsen, the top chess player in the world. If Carlsen can’t beat a computer at chess, will human dermatologists continue to outdo AI at diagnosis and treatment? I doubt it.

How do you see this interaction playing out?

Dr. Feldman: Humans are still going to want human interaction, at least for the near term, but I believe that can be digitized, too. People with a rash or lesion may soon be able to put on a virtual reality headset and choose or create an avatar of their ideal physician, whether male, female, or nonbinary. They can choose language and tone of voice and even shake hands through the headset as a virtual hand reaches out. That digital doctor can give the patient a diagnosis and treatment plan, and, now that I think of it, can probably follow up on the treatment plan at more regular intervals than we do now, all at lower cost and greater convenience. Perhaps if Libertarians had their way, the AI could send the prescription to an Amazon pharmacy for same-day delivery by drone. I guess we live dermatologists don’t stand a chance!

May this type of technology solve other problems?

Dr. Feldman: Adherence to treatment is a big issue in dermatology. The Perfect AI technology may help improve adherence. It’s not just that it gives doctors a score; a patient can become more invested in their treatments, almost like a game to see how much they can improve over time. Another app from Sensal called MyAide can precisely capture dosage time, frequency, and amount of medication dispensed, which can also help us understand our patient’s compliance and take steps to improve it.

Dr. Feldman received grant funding from Perfect Corp and is a scientific advisor for Sensal.