Competitive landscape
Dermatology is one of the most active arenas in clinical AI. Investment, regulatory milestones and consolidation have concentrated in two well-served areas, skin-cancer detection and consumer self-checks, while the broadest clinical need has stayed largely open: measuring and following disease across the full range of conditions a dermatologist sees every day. Legit.Health was built for that space.
Legit.Health is the multi-condition clinical intelligence layer for dermatology: it diagnoses across 300+ conditions and adds quantitative severity scoring, triage and monitoring, embedded in the clinical workflow and serving care providers, pharma and insurers from a single platform.
The field is crowded where the decision is narrow (one cancer, one lesion, one disease) and open where it is broad. The broad clinical layer is the position we lead.
How we compare
The real differentiator is the combination: diagnosis across 300+ conditions, quantitative severity scoring, and delivery inside the clinician's workflow. Competitors lead on parts of this (Nia Health scores severity for one disease, VisualDx spans many conditions as a reference), but none combines all of it. The table below shows a representative player from each part of the field.
| Player | Severity scoring | Diagnosis (300+) | Medical device | Clinician-delivered (not D2C) | Autonomous diagnosis |
|---|---|---|---|---|---|
| Legit.Health | ✓ | ✓ | ✓ | ✓ | ✗ |
| Skin Analytics UK | ✗ | ✗ | ✓ | ✓ | ✓ |
| DermaSensor US | ✗ | ✗ | ✓ | ✓ | ✗ |
| Nia Health DE | ✓ | ✗ | ✓ | ✗ | ✗ |
| VisualDx US | ✗ | ✓ | ✗ | ✓ | ✗ |
| SkinVision NL, consumer | ✗ | ✗ | ✓ | ✗ | ✗ |
The final column runs the other way: autonomous diagnosis is Skin Analytics' territory (CE Class III, NHS), a capability we deliberately leave with the clinician rather than replace.
The same comparison as a map: diagnostic breadth (horizontal) against severity measurement (vertical). Only Legit.Health sits top-right, combining diagnosis across 300+ conditions with quantitative severity measurement. Nia Health measures severity but for one condition; VisualDx spans many conditions but as reference, without severity measurement.
How the field is structured
The dermatology-AI market divides cleanly into four archetypes. Each is a strong business in its niche, and each is defined by a narrow scope: a single decision, a single condition, or a reference role. None spans the full breadth of clinical practice.
The market opportunity
Analyst estimates for "AI in dermatology" vary by more than an order of magnitude, because each firm draws the boundary differently (software only, or software plus imaging hardware). Rather than lean on one headline number, we anchor the opportunity bottom-up, on the buyers we actually sell to.
Our serviceable market is the intersection of three buyer segments, sized on real anchors rather than on the headline TAM:
- Care providers. An estimated 30,000 to 40,000 dermatologists across Europe form the seat floor, and the far larger pool of primary-care and teledermatology triage points that image skin is the real multiplier (the US alone sees roughly 44M dermatology visits a year). Severity and triage extend well beyond the specialist, so dermatologist headcount understates the reach.
- Pharma trials (highest value per account). Every psoriasis, atopic dermatitis, acne and hidradenitis trial needs objective, reproducible severity scoring (PASI, EASI, SCORAD, IGA, IHS4), and there are hundreds of active trials in these indications at any time. Our addressable revenue here is a per-trial endpoint line item, priced by scope, not a share of the drug sales those trials support; the large patient populations (psoriasis ~43M, atopic dermatitis ~204M, acne ~231M) and the ~$22-50B therapy area are demand context, not revenue we capture. In this segment our real alternative is human central-reader panels and imaging core labs (the CRO model); automated scoring is faster, more reproducible and auditable.
- Insurers and payers (optionality). Cost avoidance from fewer unnecessary referrals and better triage. We treat this as expansion upside, not a near-term headline.
We do not publish a single SAM figure: endpoint and seat pricing varies by trial scope and contract. The size of the opportunity is driven by the volumes above (active trials in these indications, and the addressable base of dermatologists and care sites), which is what we anchor on rather than a headline number.
Why now
- Regulators are accepting AI inside the clinical decision (the first autonomous skin-cancer AI is now CE Class III and cleared for NHS use), which de-risks credible, evidence-backed clinical AI for buyers.
- Pharma increasingly requires objective, inter-rater-reliable endpoints as inflammatory-disease pipelines expand, exactly what quantitative severity scoring provides.
- Teledermatology and remote monitoring are growing double digits, and every remote encounter needs a measurement layer.
Why the position is defensible
- Breadth compounds. A single platform spanning 300+ conditions accumulates labelled clinical data that a single-decision tool never gathers (256K+ reports processed to date, under patient consent and provider agreements). Broadening a narrow tool into a full clinical layer means repeating that data and validation work condition by condition.
- Three-sided demand from one engine. The same severity and triage output monetises across care providers (triage, follow-up), pharma (objective trial endpoints) and insurers (screening, cost control). No competitor in the archetypes above spans all three buyers.
- Workflow-embedded, no hardware lock-in. Legit.Health is software that plugs into electronic records, teledermatology and trial pipelines, and deploys without capital equipment. Device and imaging players carry a hardware sales cycle and an install base to defend; we scale as SaaS. The model is recurring software revenue (per-site and per-use), not one-off device capex or reimbursement-dependent per-procedure billing, so it compounds and carries software gross margins.
- A regulatory posture fit for purpose. MDD Class I today, with MDR Class IIb (Rule 11) certification in progress with BSI; ISO 13485 certified by BSI; ANVISA Class II in Brazil; ENS Alto for the Spanish public sector. Keeping the clinician in the loop rather than pursuing autonomous diagnosis is a deliberate go-to-market choice (lower liability, faster adoption inside clinical teams); completing the MDR IIb transition is the near-term regulatory milestone.
- A clinical-evidence engine. Independent, peer-reviewed validation underpins the platform, including non-melanoma skin-cancer detection at AUC 0.93 and severity scoring validated against clinician consensus. See Clinical evidence and Recognition.
Why it is hard to replicate
The gap is not a single feature; it is an accumulation that takes years and three kinds of trust to build.
- Regulatory time, condition by condition. Every algorithm and condition carries its own clinical validation and regulatory evidence. Breadth across 300+ conditions is not something an incumbent ships in a quarter; it is years of regulated validation. See Medical device certification.
- A clinical-evidence engine with a multi-year lead. Peer-reviewed validation accrues at a pace few match, and evidence, not code, is what unlocks hospital, pharma and payer budgets. A new entrant faces years of catch-up to reach comparable proof.
- Workflow and data network effects. Embedded across providers, pharma trials and insurers, the platform builds labelled, longitudinal data across the whole field that a single-decision tool never sees. Because it is a regulated device, that data compounds through a governed validation and change-control pathway, a real barrier to entry rather than an automatic flywheel.
- On foundation models. The real question is not ChatGPT but purpose-built medical models (Google's Derm Foundation, Med-Gemini and similar). Two things separate a model from a product here. First, the hard asset is not the model weights but a regulated, consented, longitudinal clinical dataset and the validation behind each intended use; a general model has the pixels, not the labelled outcomes over time. Second, model capability has existed for years without becoming a deployed clinical product: Google has shipped dermatology AI since 2021 and research models since, yet none is a certified, multi-condition clinical severity device running in the workflow. The barrier is certification, validation and clinical integration, measured in years, not the ability to read an image.
Proof: the most demanding buyers already chose us
Positioning is an argument; what sophisticated buyers actually do is evidence. The organisations that could adopt any tool in the field have selected Legit.Health, and they renew.
- Pharma: Johnson & Johnson (a €685K multi-year psoriasis Phase 3 programme), Almirall, Boehringer Ingelheim, Eli Lilly and Novartis use the platform for objective, reproducible trial endpoints.
- Insurers: Sanitas (Bupa) and Lux Med (Bupa Poland) deploy it for triage and screening.
- Health systems: Ribera, Hospital de Torrejón, SESPA and Hospital de Palamós run it inside clinical pathways.
Zero pharma churn is the tell: the most rigorous buyers in healthcare adopt the platform and renew. We are active in 7 markets (ES, PT, PL, BR, US, UK, DE), strongest across Europe and Brazil, where the MDR trajectory, ENS Alto and ANVISA clearances travel; the US is an expansion market where we compete on evidence and multi-condition breadth rather than an incumbent install base. See Commercial metrics and Customers by segment.
A note on strong players
Several companies genuinely lead their niches: Skin Analytics in autonomous skin-cancer screening (the first Class III CE-marked autonomous skin-cancer AI, conditionally recommended for NHS use), FotoFinder in dermoscopy and total-body imaging, and VisualDx in visual reference. We treat these as largely complementary to a multi-condition severity and triage layer rather than as substitutes for it: imaging hardware and screening pathways generate the images and cases that a clinical-intelligence layer then measures and follows over time. We see integration and partnership surfaces here as much as competition.
Recent market signals
The category is validating and consolidating, and the activity clusters in exactly the areas that are not our game, which reinforces where the open space sits.
- Autonomous AI is entering the clinical loop. Skin Analytics' DERM became the first Class III CE-marked autonomous skin-cancer AI, and NHS England issued guidance permitting autonomous use inside NHS pathways (December 2025). Regulators are now accepting AI in the clinical decision, a tailwind for credible, evidence-backed clinical AI.
- AI reached primary care. DermaSensor's FDA De Novo authorisation (2024) opened the non-specialist, point-of-care channel for skin-cancer tools.
- Capital and consolidation in imaging. A 2025 private-equity buyout of the dermoscopy-imaging leader, plus new venture-backed robotic imaging entrants, show sustained appetite for skin-imaging assets.
- Big-diagnostics money in adjacent AI. Roche's 2026 agreement to acquire the digital-pathology AI company PathAI (up to roughly $1B) signals strategic-acquirer appetite for clinical-AI platforms.
- The takeaway. Investment and clearances concentrate in cancer detection, imaging hardware and consumer risk checks. The multi-condition clinical-intelligence layer, severity and triage across the whole practice, remains the least-contested and most defensible position, and it is the one Legit.Health occupies.