A prospective diagnostic accuracy study published in May 2026 evaluated SkinVision, one of the most widely used AI-powered skin cancer risk-assessment apps, in a user-representative cohort of patients consulting doctors for lesions of concern. The app, which lets users photograph a mole or skin lesion and receive an AI-generated risk assessment, has been marketed as clinically validated with sensitivity reported between 87% and 95% and specificity between 70% and 78% in earlier evaluations — figures the new study set out to test under real-world conditions rather than curated ones.
What the study actually measured
Rather than testing the algorithm on a clean, pre-selected image set, researchers evaluated diagnostic accuracy alongside the effect of photographic conditions — lighting, angle, image quality — that ordinary users, not trained photographers, produce when scanning their own skin. That distinction matters because an algorithm’s published sensitivity and specificity figures are typically generated under more controlled conditions than a person photographing their own back in a bathroom mirror.
A crowded, growing category
SkinVision competes with apps like SkinScreener, a CE-certified Class IIa medical device under EU MDR regulation that has logged more than 300,000 downloads across 17 or more countries and is designed to recognize both pigmented and unpigmented skin cancers, including melanoma. Commercial companies have introduced multiple direct-to-consumer AI algorithms in this space, letting ordinary users self-assess suspicious lesions without a dermatologist visit, a market driven by rising skin cancer incidence and limited dermatology appointment availability in many regions.
The core tension: access versus missed diagnoses
Proponents argue these apps expand access to skin cancer screening for people who might otherwise never have a suspicious mole examined at all, particularly in areas with long waits for dermatology appointments, and that flagging even a portion of concerning lesions for follow-up care is a net public health benefit. Critics, reflected in coverage highlighting that AI-based smartphone apps assessing moles are missing skin cancer diagnoses, counter that a missed melanoma carries life-threatening consequences, and that users may develop false reassurance from a low-risk score and delay seeking an in-person exam they otherwise would have pursued.
What clinicians recommend
Dermatology researchers involved in evaluating these tools generally recommend they be used as an adjunct that encourages people to seek care sooner, never as a replacement for professional skin examination, particularly for anyone with risk factors such as a family history of melanoma, significant sun exposure, or a changing mole. Several trials, including the SPOT-study randomized controlled trial examining the app’s real-world impact on the healthcare system, are ongoing to determine whether these tools genuinely improve early detection rates or simply shift the burden of triage onto patients without improving outcomes.
What’s next
As more rigorous, user-representative studies like the May 2026 SkinVision evaluation are published, regulators and clinical guideline bodies are expected to issue clearer recommendations on how these apps should be marketed and used, particularly around disclaimers about photographic quality and appropriate follow-up thresholds. The direction of travel favors continued growth of the category, but with mounting pressure for transparency about real-world accuracy rather than best-case clinical trial figures.
