Teledermatology — the delivery of dermatological care through digital image submission and remote evaluation — has expanded access to dermatological expertise dramatically: in the U.S., patients wait an average of 32 days for a dermatology appointment (rural patients often 90+ days), while teledermatology platforms reduce wait times to 24–72 hours. Layered on top of teledermatology workflows, AI dermoscopy systems are now achieving sensitivity and specificity for melanoma detection that equals or approaches dermatologist performance in controlled validation studies — with implications for both clinical screening and patient-facing consumer applications.
AI Dermoscopy: The Evidence for Melanoma Detection
A landmark 2018 Annals of Oncology study pitting a CNN (convolutional neural network) against 58 international dermatologists in dermoscopic melanoma classification found the AI system significantly outperformed dermatologists in sensitivity (86.6% vs 74.4%) with comparable specificity — generating significant media attention and motivating regulatory response. FDA has cleared several AI dermoscopy systems: DermView, DERM AI, and SkinVision have achieved Breakthrough Device Designation or 510(k) clearance for benign/malignant skin lesion triage. The clinical caveat: these AI systems were trained and validated on high-quality dermoscopic images — performance degrades significantly with smartphone photos of variable quality, unusual lesion presentations, and skin of color (underrepresented in training datasets). AI dermoscopy is appropriate as a triage support tool, not a replacement for clinical dermatologist assessment for biopsy decision-making.
Appropriate Use in Clinical Practice
Teledermatology with AI support has demonstrated clinical value for: primary care physician referral triage (identifying which lesions require urgent dermatologist assessment vs routine follow-up); rural and underserved population access (dramatically reducing wait times and identifying conditions requiring in-person care); acne, psoriasis, and eczema management (well-validated for common inflammatory conditions where visual diagnosis is highly reliable). Limitations requiring in-person care: full-body skin examination (AI systems analyze individual lesions, not comprehensive assessment); lesions requiring dermoscopy (consumer photos inadequate); any lesion with clinical concern regardless of AI output; pediatric skin conditions; and conditions where touch (palpation) or patient history is diagnostically critical. For clinical dermatology practices using teledermatology platforms, our skin care catalog supports treatment protocols for virtually all commonly managed skin conditions.
