A New Meta-Analysis on AI and Vitiligo Arrives Without the Data That Would Prove It Works for Everyone

A systematic review and meta-analysis titled "Artificial Intelligence Techniques for Diagnosing and Quantifying Vitiligo" went up on ScienceDirect on October 5, 2026, ahead of print in the Journal of Investigative Dermatology. The authors are Nneka Ede, Fatima Khan, Imelda Vetter, Camryn Schroeder and Ammar M. Ahmed. The paper had not yet surfaced in PubMed or other indexes at the time of writing, so its pooled accuracy figures, study count and author affiliations remain unconfirmed beyond the posted abstract.

The review arrives after a run of narrower studies on the same problem. Tanvir and colleagues published a 2024 review in BioMed Research International that screened 244 records and kept 10 studies on machine learning for vitiligo detection. Kallipolitis and colleagues published a broader 2025 review in BMC Medical Informatics and Decision Making covering image analysis across vitiligo, alopecia areata and dermatitis together. A 2025 piece in Frontiers in Medicine by Mazzetto and colleagues argued for automated image analysis to replace the standard VASI and VETFa scoring tools, though that piece is an opinion article, not a review of trial evidence.

Diagnosis gets most of the attention in AI dermatology coverage, but quantification is the harder problem for vitiligo specifically. VASI and VETFa, the scoring tools dermatologists use to track lesion area and severity, rely on visual estimation of irregularly shaped, depigmented patches. Multiple papers in this space cite that subjectivity as their starting justification for building automated tools. A few products already attempt this: Vitil-IA, described in the British Journal of Dermatology in 2023, measures facial vitiligo from standardized UV photography. A platform developed with Hamzavi Dermatology computes an automated "I-VASI" score using fingertip-unit calibration. A hybrid deep-learning model published in Annals of Translational Medicine handles detection and severity scoring together.

None of the vitiligo-specific papers reviewed address how these tools perform across different skin tones. That is a notable gap, because vitiligo diagnosis depends entirely on the contrast between depigmented and pigmented skin, a contrast that varies by baseline skin tone. Broader dermatology-AI research gives reason for caution here. Work on the Diverse Dermatology Images dataset found ROC-AUC drops of 29 to 40 percent for leading dermatology AI systems on dark-skin and uncommon-disease images. A separate audit found that only 69 percent of Fitzpatrick 17k dataset labels held up under dermatologist re-review. A 2025 paper at the International Conference on Computational Science reported that Fitzpatrick types V and VI remain severely underrepresented across the datasets it examined. Whether the JID meta-analysis reports subgroup performance by skin type, or says anything about it at all, is the first thing worth checking once the full text is available.

The disease burden gives the quantification problem its stakes. A 2024 modelling study in Lancet Public Health estimated lifetime physician-diagnosed vitiligo prevalence at about 0.36 percent globally, roughly 28.5 million people, with regional estimates for Africa and the Middle East ranging from 0.18 to 5.30 percent because of sparse underlying data. A systematic review of 168 studies found depression rates reaching 62.3 percent and anxiety rates reaching 67.9 percent among vitiligo patients studied, with some cohorts reporting stigma in 100 percent of participants. A 2026 cross-cultural review covered by Dermatology Times found that in parts of South Asia, Africa and the Middle East, vitiligo is still attributed to supernatural or moral causes, or believed to be contagious.

Treatment economics add another reason the measurement question matters now. The FDA approved Opzelura, ruxolitinib cream, in July 2022 as the first repigmentation therapy for nonsegmental vitiligo, with its pivotal trials using F-VASI75 as the primary endpoint. Real-world data presented at the 2026 American Academy of Dermatology meeting showed continued efficacy alongside gaps in patient education and adherence, according to AJMC's coverage. More topical JAK inhibitors are in development behind Opzelura. A standardized, automated VASI score would serve as a trial endpoint for those drugs and support remote monitoring between visits, which makes the review's subject relevant to drug development, not just to diagnostic convenience.

The market already has a preview of what happens when accuracy claims outrun published validation. A September 2024 partnership between Skinopathy and the Vitiligo Research Foundation launched a selfie-based app that claims to identify vitiligo type and track progression with 95 percent accuracy. That figure appears in the partnership's announcement. No independent validation study for it turned up in a search of the published literature. The forthcoming JID meta-analysis will put pooled numbers like that one in the same frame as peer-reviewed, externally validated studies, and the gap between the two kinds of claim, vendor-reported versus independently tested, is where this review will earn its relevance or lose it.