Clinical evidence
This section compiles the evidence behind the automated vitiligo scoring provided by Legit.Health: the validation of the model that measures depigmentation, the image quality gate, the conference presentations and the foundational references for the VASI score. Together they document the scientific basis of the automated VASI and what remains to be measured before it is used as a clinical trial endpoint.
Validation status and reproducibility
The evidence today is the validation of the depigmentation segmentation model under the Quality Management System, reported below. The agreement of the total automated VASI with the VASI assigned by dermatologists on the same patients has not yet been reported.
Severity scoring has no objective gold standard, and VASI is no exception: it depends on how each rater estimates the affected surface in palm units and grades the remaining pigment by eye. The model is therefore measured against a pixel-level reference: the consensus of several dermatologists who traced the boundaries of depigmentation on each photograph. Matching that consensus is the realistic performance ceiling for any rater, human or AI.
Two properties are decisive for endpoint use in a trial:
- Reproducibility: scoring is deterministic. The same image yields the same depigmentation percentage at every site and every visit, with no calibration drift and no inter-reader variability in the automated read.
- Objective extent measurement: the extent of depigmentation is the core of the VASI, and the platform measures it by pixel-level segmentation rather than estimating it in palm units.
Legit.Health is a Class IIb medical device certified by BSI, Notified Body 2797, under the EU Medical Device Regulation (MDR 2017/745), meaning a Notified Body has independently assessed it against the regulation's safety and performance requirements before it could carry the CE mark. Beyond the EU, it is registered with the MHRA in the United Kingdom and with ANVISA in Brazil. Real-world performance is monitored continuously through the manufacturer's post-market surveillance and post-market clinical follow-up (PMCF) programme under MDR Annex XIV.
Legit.Health is a clinical decision support device: the automated scores provide diagnostic support and do not replace the healthcare professional's assessment.
Automated VASI
The automated VASI combines two steps. A deep learning model segments each photograph into normal skin, depigmented skin and background, and computes the percentage of depigmentation within the skin of each view. Each view is then weighted by its share of the body surface and the views are added up into the score. The second step is a fixed formula, so the validation concentrates on the first.
Production model performance
Each model is validated against its reference under the Quality Management System (IEC 62304 and ISO 14971). The depigmentation model was trained on images of vitiligo and other disorders of hypopigmentation, in which dermatologists traced the boundaries of the affected skin and their tracings were unified into a consensus mask. The training data was balanced across Fitzpatrick skin phototypes. The model was evaluated on an independent test set held out from training, using the intersection over union (IoU) of its mask with the consensus mask: the share of the two masks that overlaps, from 0 (no overlap) to 1 (identical masks).
| Model | Metric | Test samples | Result (95% CI) | Acceptance criterion |
|---|---|---|---|---|
| Depigmentation segmentation | IoU | 194 | 0.71 (0.69 to 0.74) | ≥ 0.69 |
The criterion was set from the published literature and from the variability between dermatologists tracing the same images, and the model meets it.
Performance across skin phototypes
Contrast between pigmented and depigmented skin is the main visual cue for the model, and that contrast is lowest on the lightest skin. The model's IoU was analysed separately for each group of Fitzpatrick skin phototypes:
| Skin phototype | Test samples | IoU (95% CI) |
|---|---|---|
| I to II | 64 | 0.69 (0.63 to 0.74) |
| III to IV | 93 | 0.72 (0.68 to 0.76) |
| V to VI | 37 | 0.74 (0.69 to 0.78) |
The results are similar across the three groups, and their confidence intervals overlap. In every group the model meets the criterion set from the published literature and from the variability between dermatologists.
Variation in capture conditions
During training, the images were varied by rotation, mirroring, cropping, zoom, blur, and changes of colour, brightness and contrast, tuned so as not to change the clinical appearance of the skin. The variation teaches the model the appearance of depigmentation rather than the conditions of one camera or room, which matters in a multi-site trial where sites use different phones, rooms and lighting. Image quality at capture is checked by DIQA, below.
Dermatology Image Quality Assessment (DIQA)
Dermatology Image Quality Assessment (DIQA): Artificial intelligence to ensure the clinical utility of images for remote consultations and clinical trials. Hernández Montilla I, Mac Carthy T, Aguilar A, et al. Journal of the American Academy of Dermatology. 2023;88(4):927–928. DOI: 10.1016/j.jaad.2022.11.002 | PMID: 36526082
Reliable VASI scoring depends on the quality of the input photographs. DIQA is the image quality assessment algorithm that acts as a quality gate in the vitiligo imaging workflow, checking every image against consistent quality criteria across investigator sites before it reaches the segmentation model. The dependency is sharp in vitiligo, where each visit needs several views of the body and flash glare can make normally pigmented skin look depigmented.
Conference presentations
AAD Annual Meeting 2026, Denver
Oral presentation: Artificial intelligence-based quantification of the Vitiligo Area Scoring Index
Presented on 27 March 2026. New clinical research on the Automatic Vitiligo Area Scoring Index (AVASI), addressing the variability and subjectivity of manual VASI scoring.
EADV Congress 2026, Vienna
E-poster: Automatic Vitiligo Area Scoring Index (AVASI)
The design of the AVASI study, a prospective, observational, multicentre study that compares the VASI calculated automatically from the photographs with the VASI the treating dermatologist scores at the visit and with the VASI an independent expert assigns on the same images.
Foundational VASI references
The platform does not introduce a new, unvalidated scale. It automates VASI, the measure of vitiligo extent that clinical trials already use, so the endpoint your protocol specifies is the endpoint the algorithm reports.
- Hamzavi I, Jain H, McLean D, Shapiro J, Zeng H, Lui H. Parametric modeling of narrowband UV-B phototherapy for vitiligo using a novel quantitative tool: the Vitiligo Area Scoring Index. Archives of Dermatology. 2004;140(6):677-683.
- Rosmarin D, Passeron T, Pandya AG, et al. Two phase 3, randomized, controlled trials of ruxolitinib cream for vitiligo. New England Journal of Medicine. 2022;387(16):1445-1455.
For the full list of clinical evidence across all indications, see the clinical validation section.
Accelerate your trials
Join pharmaceutical companies, biotechs, and CROs reaching the market 4-6 months faster with superior data quality. Fill in the form to discuss your study requirements.
