This Arkangel AI whitepaper reviews what supportive AI can and cannot do today in dermatology, and what governance it needs for safe adoption by hospitals and clinical teams in Latin America. It is a narrative synthesis of 24 sources from targeted searches of PubMed, Europe PMC, Crossref, publisher pages and regulatory sources, covering four tasks: skin-lesion and skin-cancer classification, broad dermatologic diagnostic support, teledermatology and triage, and data and implementation governance. Four focused Europe PMC searches on July 22, 2026 (40 screened positions) added regional sources.
Evidence is most mature for pigmented lesions, skin cancer and dermoscopy, where studies since Esteva et al. show performance comparable to dermatologists on curated image tasks and human-AI collaboration depends on how the model is combined with readers. Broad differential diagnosis, longitudinal follow-up and people with skin of color remain less studied. The search found no prospective clinical validation of dermatology AI in Latin America; regional evidence covered teledermatology, dermoscopy training and telemedicine, such as a São Paulo teledermatology program with 30,976 people that changed referral and waiting times without evaluating AI. The authors support AI for reading support, triage, education and referral under clinical review, not autonomous diagnosis, and call for representative data, local validation across skin types and devices, and governed workflows. The document is a whitepaper, not peer reviewed and not a systematic review; all authors are affiliated with Arkangel AI, and the evidence it cites does not evaluate Arkangel products.
