Dermatology AI scribe 2026: lesion lists, biopsy tracking, Mohs surgery notes, and the photo documentation reality

May 8, 2026 · 6 min read

Dermatology visits don't fit the SOAP note paradigm cleanly. A full-body skin check generates 8-15 lesion observations in 12 minutes. A biopsy visit produces both the procedure note and a tracking record for pathology follow-up. A Mohs surgery visit has its own structured documentation requirements. Photos are integral to the workflow. General AI scribes built around chief-complaint narratives don't capture this well.

The 2026 dermatology-aware AI scribe stack handles four things general scribes miss: structured lesion lists with anatomic precision, biopsy tracking with pathology-result follow-up, Mohs procedural documentation, and photo cross-reference.

The lesion list problem

A typical full-body skin check generates output like:

• "3-mm pink macule, right cheek, 2 cm lateral to nasolabial fold — benign-appearing, follow"
• "5 × 4 mm brown lesion, left scapular region — clinically benign nevus, asymmetry minor, photograph for monitoring"
• "8-mm pearly papule with telangiectasia, right temple — clinical BCC, biopsy planned"
• ...continue for 8-15 lesions

A general scribe writes a paragraph. A derm-aware scribe writes a structured lesion list, each with anatomic location, size, morphology, clinical impression, and disposition. Each row is independently codeable, billable, and trackable across visits.

The biopsy tracking layer

Every biopsy generates two documents: the procedure note and the pending-pathology tracker. The tracker is the part general scribes drop. The derm-aware pipeline:

• Auto-creates a pending biopsy entry in the patient's chart with the lesion ID + location + procedure date
• Cross-references when pathology returns — closes the loop with the lesion record
• Triggers patient communication if pathology is abnormal (BCC / SCC / melanoma)
• Schedules follow-up automatically based on result severity

For a busy derm practice, biopsy tracking is the difference between catching a pending result and a missed melanoma diagnosis. The structured layer is what makes audit defense work.

Mohs surgery documentation

Mohs visits have their own structured note format:

• Pre-op clinical impression + photo documentation
• Stages: each Mohs stage with map, anatomic depth, frozen-section interpretation, decision (clear / take more)
• Reconstruction: defect size, repair approach, suture material, layers
• Post-op care + follow-up plan

An AI scribe that produces a structured Mohs note with stage-by-stage documentation supports CPT billing (17311 / 17312 / 17313 / 17314 stages) defensibility.

Photo cross-reference

Most modern derm practices photograph clinically-relevant lesions and store them in the EHR or a derm-specific photo system. The AI scribe should:

• Reference photos taken during the visit (timestamp + lesion ID)
• Cross-reference prior photos for monitored lesions (size comparison, morphology change)
• Flag visually-suspicious lesions where the clinician's verbal description and the prior photo trajectory diverge

The dermatology-aware system prompt

You are documenting a dermatology encounter.

INPUT:
- Encounter audio transcript (verbatim)
- Visit type: full-body skin check / problem-focused / procedure / Mohs / cosmetic
- Patient profile: skin type, prior biopsy history, prior dx (melanoma in family, etc.)

OUTPUT depends on visit type:

For full-body skin check or lesion-focused visits:
- LESION LIST (structured): for each lesion mentioned, output:
  * Lesion ID (sequential within visit)
  * Anatomic location (precise: not "back" but "right scapular region")
  * Size (in mm if mentioned)
  * Morphology (papule / macule / plaque / nodule, color, shape, surface)
  * Clinical impression (benign nevus / dysplastic / BCC / SCC / melanoma / actinic keratosis / etc.)
  * Disposition (observation / photograph for monitoring / biopsy / cryosurgery / etc.)
  * If biopsy: technique (shave / punch / excision), expected pathology question
- Patient education content
- Sun protection counseling content if discussed

For procedure or Mohs visits, use procedure-specific schema (request separately).

For each clinical fact, cite the transcript line. Flag any lesion description that's ambiguous or under-specified.

Vendor matrix — dermatology AI scribes 2026

For solo derm practices on EMA, native gIE is the natural fit. For multi-clinician practices wanting custom workflow integration with their specific photo system, the DIY path with FHIR API and photo cross-reference logic outperforms.

The cosmetic / aesthetic angle

Cosmetic / aesthetic dermatology has different documentation needs:

• Pre-treatment photos with measurement reference
• Treatment area mapping (units of toxin per zone, filler volume per area)
• Touch-up plans
• Adverse event tracking
• Patient consent for cosmetic procedures (separate from clinical)

An AI scribe that handles both clinical and aesthetic visit types in one pipeline serves the typical mixed-practice well. Different prompt for visit type, same audio capture, same transcription pipeline.

BAA chain

For DIY: practice + EHR vendor + transcription vendor + LLM vendor + photo system vendor. 5-party chain when photos are involved.

When to start

If your derm practice does more than 30 full-body skin checks per week or has a Mohs surgery component, structured lesion-list and procedural documentation pay off in audit defensibility, billing accuracy, and pending-result tracking. The DIY stack is buildable in 4-6 weeks of clinician + IT time, with the lesion-list schema being the heaviest part.