AI in Dermatology Market (2026-2036)

The growth of this market is mainly driven by the modest accuracy of clinician visual examination, which detects melanoma with sensitivity of 40% to 70% and therefore gives a cleared second-read device a quantified improvement to demonstrate, by a dedicated class II classification that places a cleared product in the hands of physicians without skin cancer training and therefore in the primary care setting, and by severity scoring in inflammatory disease, where an area and severity index assessed by eye varies between assessors and a reproducible measurement governs access to biologic therapy. Dermatology is also the specialty whose primary data is a photograph, which costs almost nothing to acquire and transmit, so the cost of getting data to a model is lower here than in any imaging-based specialty.

However, factors such as the adjunctive and post-identification scope of the cleared claim, the restriction of that claim to physicians not trained in skin cancer diagnosis and management, the absence of any payment attaching to the software itself, and the separation between the practice that buys the device and the payer that captures the avoided referral are expected to restrain the growth of this market to some extent. A cleared product cannot be sold as a screening tool or as a substitute for specialist assessment, and a primary care visit is reimbursed identically whether a software second read was performed, so adoption depends on a health system or payer purchasing centrally on behalf of practices that receive none of the saving.

21 CFR 878.1830, the class II classification for a software-aided adjunctive diagnostic device for use by physicians on lesions suspicious for skin cancer, product code QZS, cleared through the 510(k) pathway and reviewed by the General and Plastic Surgery panel, defined as a prescription device that uses a software algorithm to analyse optical or other physical properties of a skin lesion and classify it, intended as an adjunctive second-read tool for physicians not trained in skin cancer diagnosis and management after a suspicious lesion has been identified, and not a standalone diagnostic that can confirm a clinical diagnosis sets the limits of what this market is permitted to sell, and those limits are easily missed because the technology discussion in dermatology artificial intelligence is usually framed around accuracy. The device is adjunctive and cannot confirm a diagnosis, so it cannot replace a biopsy or a specialist opinion, it operates only after a suspicious lesion has been identified, so it does not find lesions nobody noticed and is not a screening product, and it is intended for physicians not trained in skin cancer diagnosis and management, which places the buyer in primary care and urgent care rather than in dermatology. The addressable market is therefore the general practice encounter in which a patient points at a mole, not the dermatology consultation and not a population screening programme.

The U.S. Preventive Services Task Force recommendation on screening for skin cancer, dated 18 April 2023, which is an I statement concluding that the evidence is insufficient to assess the balance of benefits and harms of visual skin examination by a clinician in asymptomatic adolescents and adults, excluding those with a personal or family history of skin cancer or with symptoms removes the screening route in any case. There is no recommendation to examine asymptomatic skin, so no quality measure obliges a health system to run a skin cancer screening programme and no budget exists to be captured by software that would support one. What the same recommendation does supply is a measured baseline, since the same recommendation, which reports that visual skin examination has modest accuracy for detecting melanoma, with sensitivity of 40% to 70% and specificity of 86% to 98%, that the evidence that visual skin examination reduces skin cancer morbidity or mortality is inadequate, and that the German screening programme studies reviewed were nonrandomised and limited in applicability to the United States. A clinician examining a lesion by eye therefore misses between three and six melanomas in ten, and that figure is the benchmark a software second read is compared against rather than the performance of a dermatologist or of histopathology, which is an unusually precise statement of the standard of care a product in this field is displacing.

A primary care physician facing an uncertain lesion refers, and the referral consumes a dermatology appointment in a specialty with constrained capacity in most health systems, so the economic value of a second read is the referral it prevents and the appointment it releases for a patient who needs it. That value accrues to the payer or to the integrated health system rather than to the practice performing the examination, which is why the organisations buying at scale in this market are health systems, payers and primary care networks operating under capitation rather than individual practices. A practice paid the same for the visit whether or not it used software has no financial reason to buy, which is the single largest obstacle to adoption at the point of care.

The Centers for Medicare & Medicaid Services calendar year 2026 hospital outpatient prospective payment system final rule, published on 21 November 2025, in which the agency solicited comments from the public on payment policies for software as a service under the outpatient system, stated that it received many comments on future payment methodologies, finalised no payment policy and said it will consider the comments received for future rulemaking, so no settled mechanism exists for paying for software of this kind as a service. The consequence in dermatology is that revenue concentrates where a procedure or a service already carries payment, namely dermatopathology, where the slide is already being read and billed, teledermatology, where the consultation is already reimbursed in most systems, and clinical trial scoring, where a sponsor pays for a measured endpoint. Products whose benefit is an avoided referral have the strongest clinical argument and the weakest payment position in the market.

Severity in psoriasis, atopic dermatitis, acne and hidradenitis suppurativa is scored on indices assessed by eye, those scores determine eligibility for and continuation of expensive biologic therapy, and the same lesion scored by two assessors yields different numbers. A reproducible automated score therefore has value to a pharmaceutical sponsor running a trial whose endpoint is that index, to a payer administering a treatment criterion defined by it, and to a clinician documenting a continuation decision. The buyer in each case has a budget that already exists and a measurable reason to spend it, which is why severity scoring grows faster than lesion triage despite triage holding the larger share.

The European Union Artificial Intelligence Act, which entered into force on 1 August 2024 and became generally applicable on 2 August 2026, and under which high-risk artificial intelligence systems embedded in products regulated by the sectoral legislation listed in Annex I, which includes medical devices, apply from 2 August 2028 following the extension of that transition period by the AI Omnibus governs the European side alongside the Medical Device Regulation, under which software intended to inform decisions for diagnostic purposes carries a higher classification than the imaging hardware it runs on. Europe has also permitted a route the United States has not, since consumer-facing lesion assessment applications have been placed on the market there as regulated medical devices used directly by the patient, whereas the United States classification restricts use to a physician. That divergence is the reason the consumer-facing segment of this market exists in Europe and barely exists in North America, and it is a regulatory rather than a commercial difference.