EADV Review Finds Gaps in Dermatology AI Apps
Artificial intelligence is becoming increasingly common in dermatology, with smartphone applications promising everything from skin cancer screening and mole analysis to personalized skincare recommendations and teledermatology support. But according to a new review from the European Academy of Dermatology and Venereology (EADV) Artificial Intelligence Task Force, many of these tools still lack the clinical evidence and regulatory transparency needed for widespread adoption.
The review evaluated dermatology-related mobile applications available through the Apple App Store and Google Play Store across 44 European countries, focusing on how these apps are developed, validated, and regulated.
Most Apps Target Patients
Researchers found that the majority of dermatology apps were designed for patients rather than healthcare professionals. Of the applications reviewed, 60% targeted patients, while 40% were intended for clinical use.
Teledermatology represented the largest category, followed by general dermatology education, disease-specific guides, and calculators. Self-diagnosis tools were also common, with artificial intelligence incorporated into many of these applications.
Although AI continues to expand rapidly, relatively few apps reported using validated AI models supported by published clinical research.
Clinical Evidence Remains Limited
One of the review’s biggest concerns was the lack of rigorous clinical validation.
Only 20% of evaluated apps were linked to at least one PubMed-indexed publication. Just 24 applications had been studied in clinical trials, five had undergone randomized controlled trials, and only six had been evaluated through real-world post-deployment studies.
These findings suggest that while AI-powered dermatology apps continue to grow in popularity, many have not yet accumulated the evidence typically expected for clinical decision-support technologies.
Transparency and Regulation Still Lag Behind
The review also identified significant gaps in regulatory transparency.
Only 2% of reviewed applications reported having a CE (Conformité Européenne) medical device marking, while just 10% stated compliance with the General Data Protection Regulation (GDPR).
Researchers also noted that many developers provided little publicly available information regarding AI validation, algorithm development, or performance testing. This lack of transparency makes it difficult for clinicians and patients to independently evaluate how these systems were developed and whether they perform consistently across different populations.
The review further highlighted concerns regarding skin tone representation, with limited evidence that many AI systems had been adequately validated across diverse skin types.
Recommendations for Safer AI in Dermatology
Despite these limitations, the authors emphasized that dermatology AI applications still have significant potential to improve patient education, teledermatology, and skin cancer screening.
Rather than discouraging innovation, the review recommends stronger clinical validation, improved regulatory oversight, greater transparency regarding AI development, and broader testing across diverse patient populations before these tools are relied upon more broadly in clinical practice.
As AI continues to become part of everyday dermatology, ensuring these technologies are accurate, transparent, and supported by high-quality evidence will be essential for both patient safety and clinician confidence.
