Methodology & evidence

A care score, not an attractiveness score

The 0–100 scores rate how well skin and hair are cared for: hydration, texture, tone evenness, hair condition, grooming. They never rate attractiveness, never compare anyone to an ideal or a celebrity, and never comment on ethnicity, weight or facial features. A well-groomed person typically lands at 70–85; scores above 90 are reserved for exceptional care.

Capture quality decides what we claim

Every photo is graded for lighting, angle and sharpness. If a photo isn't comparable, we withhold the score comparison — a false “no progress” is worse than no number. Differences within ±2 points are shown neutrally as measurement noise: two photos of the same face can differ from light and angle alone.

Honest comparison over time

On a re-scan the model sees both photos side by side and may only report visible differences. “Unchanged” and “regressed” are first-class answers — the prompt explicitly forbids inventing improvement to please you. When differences are plausibly caused by lighting or angle, the comparison's stated confidence drops instead of masquerading as change.

Four levels of evidence

Every plan step carries an evidence level.

Strong

Multiple randomized trials or meta-analyses.

Moderate

A few solid studies.

Weak

A single small study.

Anecdotal

Clinical practice without formal studies. Haircut suitability is one example.

Citations only from a curated library

The AI model is not allowed to “remember” a study — it may cite exclusively from the curated library of verified publications below. If no library entry supports a recommendation, the step is labeled anecdotal with no source. The server additionally checks every citation against the library, so a fabricated reference has no way to reach you.

AI “after” images are illustrations

The “after” simulation shows a possible direction with consistent care over 6–12 months. It preserves identity, features and skin tone and stays within realistic outcomes — it is not a prediction or a promise. We never simulate surgical changes.

Limits of the tool

MiraSkinLab is an educational tool, not a medical device. Clinical metrics (pore size in µm, hydration %, wrinkle depth) cannot be measured from a single selfie — so we don't claim them. We provide no diagnoses or drug dosing, and for anything that needs a professional we point you to a licensed dermatologist or aesthetic medicine doctor.