2026: certified for the whole of the skin.
This year Legit.Health became the first medical device certified under the EU Medical Device Regulation to support the diagnosis of every visible skin condition and to measure how severe it is. Around that certificate sit five more achievements: our information system certified at the highest category of the Spanish National Security Framework, a new device version compared with its predecessor on the same pigmented lesions, clinical studies showing that clinicians diagnose better with it, an app with an AI assistant, and new indexes in the API.
Regulation
The first MDR certificate for every visible skin condition.
On 15 September 2026, BSI certified Legit.Health under the EU Medical Device Regulation as a Class IIb medical device. Its certified intended purpose does not name a single disease: it covers diagnosis support across the whole of the skin, its appendages and the associated mucous membranes, and the measurement of how severe a condition is. No other company holds an MDR certificate for that scope. It is also registered with ANVISA in Brazil and with the MHRA in the United Kingdom.
How it compares
Most certified dermatology AI is certified for skin cancer only. Set side by side with DERM by Skin Analytics, as each manufacturer states it in its own instructions for use, the difference is in who and what each device is for:
- Conditions: Legit.Health covers every disease of the skin; DERM covers lesions suspicious for skin cancer and excludes rashes, eczema, infections and lupus.
- Severity and follow-up: Legit.Health measures severity at each visit, so the response to treatment can be followed; DERM lists monitoring treatment and staging disease among its limitations.
- Children: Legit.Health is intended for adults and children; DERM is for patients aged 18 or over.
- Skin types: Legit.Health is intended for Fitzpatrick types I to VI; DERM was evaluated mostly on types I to IV and is to be used with caution on the others.
Legit.Health is intended for all six Fitzpatrick phototypes. DERM's instructions for use ask for caution outside types I to IV, which in practice means patients with darker skin.
The certification page has the full comparison, the certificate itself and why a system that integrates Legit.Health does not become a medical device of its own.
Security
Our information system, certified at the highest category.
In June 2026 the information system behind Legit.Health was certified against the Spanish National Security Framework (Esquema Nacional de Seguridad, ENS) in category High, the highest of its three categories, which is what Spanish public health services require of the systems that handle their patients' data. The certificate was issued by Applus+ and covers the information system that supports the design, development and maintenance of our diagnosis-support software.
Clinical evidence
A programme of studies, each answering a different question.
Legit.Health’s clinical validation programme brings together more than ten clinical studies examining diagnosis support, malignancy assessment, severity measurement, referral decisions and use in clinical practice, across different conditions and patient populations.
This report highlights selected findings and advances from 2026. For the study catalogue and the clinical benefits each study supports, consult the Clinical Validation Studies in the Instructions for Use, the reference documentation for Legit.Health’s clinical evidence.
The device
Comparing diagnosis ranking between versions.
In IDEI_2023, we prospectively compared the two device versions in clinical cases. The new version put the correct condition first more often than the previous one, and among its first three more often too; within the first five the two were level.
| Correct condition in the ranked list | Previous version | Legit.Health Plus |
|---|---|---|
| Top-1 | 71.43% (20/28) | 78.57% (22/28) |
| Top-3 | 82.14% (23/28) | 89.29% (25/28) |
| Top-5 | 89.29% (25/28) | 89.29% (25/28) |
Distinguishing malignant from benign lesions
In the prospective arm of IDEI_2023, Legit.Health’s malignancy score achieved an AUC of 96.69% (89.00%-100.00%) for distinguishing malignant from benign pigmented lesions, assessed against histopathology. The interval shown is the 95% confidence interval. This estimate comes from a small prospective sample, with independent confirmation planned.
AUC measures how well the scores distinguish malignant from benign lesions across decision thresholds. It is not the percentage of cancers detected at a particular cutoff.
Published clinician results
Getting the first differential diagnosis right is a demanding task for clinicians too. In a teledermatology study of 963 cases, primary-care physicians achieved 44% accuracy for their first diagnosis (Top-1), and dermatologists 63%, against a dermatologist panel's reference diagnosis. Liu et al., Nature Medicine (2020).
In a cohort of 144 cases from a prospective study of suspected skin neoplasms at South Korean hospitals, attending dermatologists achieved 61.8% Top-1, 69.4% Top-2 and 71.5% Top-3 accuracy. Kim et al., PLOS ONE (2022).
These results provide context for diagnostic accuracy: conditions and diagnostic reference standards differ from IDEI_2023, which evaluated histopathology-confirmed pigmented lesions with nevus subtypes grouped together. They are not a direct comparison with Legit.Health.
Summary of metrics
Legit.Health’s evidence programme includes more than ten clinical studies alongside technical evaluations during development. The cards below highlight three measures from development testing. Clinical study results are reported separately above and in the Instructions for Use.
- ICD-11 categories in the differential
- 346
- From the common to the rare, each with its code.
- Malignancy AUC · Development test set
- 92%
- Ability to rank malignant images above benign ones across thresholds on a held-out test set. AUC is not the proportion of cancers detected at a chosen cutoff.
- Correlation with expert image-quality ratings
- 0.90
- Pearson correlation describes how scores vary together, rather than the percentage of identical ratings.
Evidence
Clinicians diagnose better with Legit.Health.
Four reader studies asked the same question: when the same clinicians read the same images of real cases, first on their own and then with Legit.Health, how much more often do they get the diagnosis right? Each one found a gain. In every study that compared the two groups, the gain was larger for primary care physicians, who are the first to see most skin conditions.
The charts show how often clinicians placed the correct diagnosis first, with and without Legit.Health. Each study reports results for its own set of clinical cases.
Each headline shows the highest result from an individual study. The specialty analysis in MAN_2025 is exploratory.
- in primary care physicians' diagnostic accuracy
- Up to +28 pp
- correct first-ranked diagnoses by primary care physicians in a reader study
- Up to 90%
- in dermatologists' diagnostic accuracy
- Up to +16 pp
Primary care physicians
BI_2024Generalised pustular psoriasis and other conditions+20 percentage points 49% to 69%PH_2024Skin lesions in primary care+18 percentage points 64% to 82%SAN_2024A wide range of skin conditions+27 percentage points 63% to 90%MAN_2025Dark skin, Fitzpatrick V and VI+28 percentage points 37% to 65%
Dermatologists
The studies
MAN_2025, dark skin: Every one of its 149 cases was a Fitzpatrick type V or VI presentation, the skin types that most datasets under-represent. With Legit.Health, readers of every specialty together, nurses included, named the right condition first in 65.07% of cases.SAN_2024, a wide range of conditions: Primary care physicians and dermatologists diagnosed cases across thirteen conditions, from the common to the rare. Read the study.BI_2024, generalised pustular psoriasis: Evaluated recognition of this rare disease alongside conditions with a similar presentation. Published in JMIR Dermatology in 2026. Read the study.PH_2024, primary care: General practitioners at a primary care centre, on the skin lesions they see every week. Read the study.
Publications
Peer-reviewed, and published this year.
Four studies were published in 2026 in peer-reviewed dermatology journals, from diagnosis support in a rare disease to automated scores of how severe a condition is.
A multi-reader study of how Legit.Health helps clinicians recognise a rare and severe form of psoriasis.
An automated severity score for generalised pustular psoriasis, calculated from clinical photographs.
The Acne Lesion And Density INdex, an automatic measure of how severe acne is.
Measuring the area of a lesion from a single photograph, without a ruler.
Image capture
A better photograph, a better result.
Legit.Health analyses the whole photograph and gives a single result, so whatever else is in the frame affects it. A poor photograph is the problem we see most often in practice, and the reason the app now helps the person taking it.
The lesion is a small mark on the forearm, among freckles, with the desk behind it.
The most common capture problems
With two lesions in the frame, Legit.Health cannot know which one the question is about. A crop around one lesion, or one photograph per lesion, asks one question at a time.
A pen mark on the skin, or a circle added afterwards with a photo editor, becomes part of what is analysed. The photograph should show the skin as it is, without marks or edits.
From far away the lesion is a few pixels among freckles, clothing or the desk. With the camera closer, the photograph shows the lesion rather than what surrounds it.
When several conditions come out about equally likely, none of them stands out. The entropy shows how uncertain the result is, and a crop around the lesion or a new photograph usually helps.
The photography guide in the instructions for use covers framing, focus and light in detail. The crop suggestion and the consistency check, described below, are how the app solves these problems as the photograph is taken.
The quality check, DIQA
DIQA (Dermatology Image Quality Assessment) is the quality check that scores each photograph from 0 to 100 before analysis. The acceptance thresholds are the same as before. Its correlation with expert ratings of image quality is in the summary of metrics above, and the features page shows the check in the app.
The app
An AI assistant beside the report, not in place of it.
The app, which customers embed in their own systems, now includes an AI assistant for the two moments where people most often need help: taking a photograph that can be assessed, and understanding the report. It is separate from the medical device. It never diagnoses, never recommends a treatment and never changes the medical result.
The screen shows the scores for an example clinical case.
What is new
It checks focus, light, framing and distance, and that the photo matches the body site, and offers a retake on the spot instead of a refused photo later.
It finds the affected skin and suggests a crop when the lesion is small or off-centre in the frame.
A short explanation inside each section of the report, for clinicians: the indexes, the differential and the severity score.
Clinicians ask about the report, for example what the top result means, and the answers stay within the report and skin health.
Also new
- Why a photograph was refused: when the device's quality check refuses a photograph, it says in plain words what to change.
- Consistency check: it looks at the photographs alongside the device's result and offers a retake only when a fixable capture problem, such as blur, glare or a lesion cut off or missing from the photograph, prevents a usable assessment.
- Dermoscopy check: it flags a photograph analysed as a dermoscopic image that shows neither a pigmented lesion nor a skin tumour.
- Under your control: each step is switched on or off from the URL, like the rest of the app, and if the assistant is unavailable the report is shown as usual. The PDF of the report never contains the assistant's text. The implementation page explains how.
The API
New indexes that help decide what to do next.
The new device returns more than a ranked differential. Malignancy and referral indexes are estimates for the case being assessed; entropy measures uncertainty in the differential. Clinicians interpret them alongside clinical history and examination; the features page shows each one in the interface.
- Entropy: how spread out the differential is. It is low when one condition clearly leads and high when several are about equally likely, so an uncertain case can go to a person instead of down an automatic pathway. New in this version.
- Referral probabilities: how likely the case needs an urgent referral, within 48 hours, and how likely it needs a high-priority one, each in low, moderate and high bands, for triage and waiting-list priority. New in this version.
- Condition present: whether the photograph shows a skin condition at all, read before anything else.
- Malignancy probability: how likely the lesion is malignant, shown as a gauge with the same bands, for fast-track pathways.
- Six safety indicators: whether the case is malignant, pre-malignant, associated with malignancy, pigmented, or needs an urgent or a high-priority referral. New in this version.
Explore an example report in the API below.
{ "identifier": "a9ae9e81-…", "imageAnalyses": [ { "technicalSummary": { "imageQuality": { "score": 7.4, "interpretation": "good" } } } ], "studyAggregate": { "findings": { "hypotheses": [ { "identifier": "psoriasis", "probability": 84.2 }, … ], "entropy": 22.1, "riskMetrics": { "urgentReferralProbability": 4.1, "highPriorityReferralProbability": 62.0, "malignantConditionProbability": 1.2 } } }}
- Quality: Whether each photograph can be assessed and how good it is, so a poor one is retaken before anything else is read.
- Diagnosis: The conditions the photographs may show, ranked by probability, and how certain that ranking is.
- Referral: How likely the case needs a specialist and how likely the lesion is malignant, to set its priority and its pathway.
- Severity: The intensity and the extent of each clinical sign, to grade the condition and follow it from one visit to the next.
Jetzt starten
KI-gestützte Dermatologie, validiert durch peer-reviewed Forschung. Vertraut von führenden Krankenhäusern in Europa. Füllen Sie das Formular aus, um zu sehen, wie unsere CE-gekennzeichnete Plattform Ihre Praxis transformieren kann.
