Skin analysis without AI diagnosis, why the line matters

At trade fairs and in vendor brochures one promise has taken hold in recent years: the scanner captures, an algorithm evaluates, and out comes an assessment that sounds like a diagnosis. For a practice this is tempting because it appears to remove work. In fact it shifts responsibility to a place that cannot carry it.

What a skin scanner actually delivers

A 3D skin scanner measures geometry and reflection. It records how deep a wrinkle runs, how much volume a region holds, how strongly an area absorbs in the UV spectrum, how red a patch appears relative to its surroundings. These are physical measurements and they are reproducible when positioning and lighting are right. So far, nothing is contested.

What a scanner does not measure is meaning. Whether redness across the cheeks is early rosacea, a reaction to a skincare product, or simply the result of exercise twenty minutes ago is not in the data. That attribution requires history, course and examination. It is medical work, not compute.

Why professional bodies warn

The German dermatologists association has repeatedly warned about devices and apps that assess skin changes automatically and hand patients a risk estimate. The criticism targets two effects: an unremarkable result can lead someone to skip an appointment they should have kept, and a flagged result creates worry and urgency where no workup was needed. Both error directions carry a cost, the first medical, the second in trust.

Where the regulatory line runs

The distinction is not only ethical, it has a legal edge. Software that supplies data from which a physician derives a diagnosis is classified differently under the MDR than software that makes a diagnostic statement itself. MDCG 2019-11 sets out the criteria, and the closer a product sits to independent assessment, the higher the classification, with everything that entails for conformity assessment.

For a practice this means: a vendor promising diagnosis-adjacent statements must be able to show the corresponding classification and conformity assessment. If neither is produced on request, that is not a detail. It is a reason to walk away.

How we handle it

The Dermalia report reproduces measurements, places them on a scale and describes what stands out. It states no diagnosis, no suspected diagnosis and no therapy. At the end of every report stands the line we consider the most important sentence in the document: discuss this report with your practice.

• One scale from 0 to 100, higher is better, with no hidden inversions between chapters.
• Findings are named, not judged: "the redness map shows elevated values across the cheek region" rather than an attribution to a condition.
• Optional text modules that generate a summary can be switched off per practice and are off by default.
• When a question contains terms suggesting an emergency, the Q&A module points straight back to the practice instead of answering.

What the practice gains

A report without a diagnosis takes nothing away from the practice, it prepares the conversation. Patients arrive with a specific question rather than a vague worry or a finished opinion from the internet. The value lies in the preparation, not in the shortcut. And legally, the judgement stays where it belongs.