Design and test a clinician-reviewed workflow for scalp or skin imaging, consultation notes and follow-up using synthetic cases.
Each teaching day runs from 9:00 a.m. to 4:30 p.m. Live virtual sessions follow Eastern Time. On-campus sessions follow local time.
Breaks are at 10:30 a.m. and 2:45 p.m. Lunch runs from 12:15 to 1:15 p.m.
Join by video with a computer for screen-sharing and guided labs. Cohort dates will be announced.
Build a capture checklist, map the vendor data flow and evaluate a model against a set of reference decisions.
| Time | Session |
|---|---|
| 9:00–10:30 a.m. | Select a task and define human responsibility |
| 10:45 a.m.–12:15 p.m. | Build a reproducible image set |
| 1:15–2:45 p.m. | Privacy, consent and vendor data flows |
| 3:00–4:30 p.m. | Evaluate image analysis against a reference |
Audit an AI-drafted note, rehearse an exception by screen sharing and decide whether the pilot should proceed.
| Time | Session |
|---|---|
| 9:00–10:30 a.m. | Review hair and skin outputs in context |
| 10:45 a.m.–12:15 p.m. | Clinical note drafting and error correction |
| 1:15–2:45 p.m. | Human review, exceptions and downtime |
| 3:00–4:30 p.m. | Pilot measurement and model changes |
Allow two additional hours to revise and submit the independent capstone using feedback from the course.
Hair restoration and aesthetic clinicians, clinical leads and practice managers responsible for imaging or documentation. Coding experience is optional.
Map one imaging or documentation task in your practice. Bring an empty note template and a list of the people who review it. Use synthetic images and notes for the workshop.
Study times are estimates for the proposed course.
Trace a consultation from intake to photographic capture, interpretation, documentation and follow-up. Separate image-quality screening, numerical measurement, note drafting and clinical decisions. Identify the cost of a missed finding and the burden of false alarms for each task.
Choose either scalp follow-up photography or aesthetic consultation notes. Specify an intended user, an allowed output, an excluded use and a manual fallback.
A one-page workflow brief naming the accountable reviewer and the point at which approval is required.
Standardize framing, focal distance, lighting, angle, hair part and camera settings where feasible. Record field of view, visit date and device version. Examine how glare, pigment variation, compression and inconsistent poses can alter apparent change.
Sort image pairs into comparable, repeat capture and exclude. For each rejected pair, give the operator a concrete correction and record why a baseline cannot be recovered.
A capture checklist, image manifest and rejection log for a simulated longitudinal case.
Follow an image and its metadata from camera to application, vendor, export and deletion. Review identifiers in filenames, faces, reflections and embedded metadata. Compare access roles, retention periods and permitted uses of images. Use the HHS guidance to frame questions for the privacy lead.
Diagram a fictional clinic sending images to an external processor. Identify which team member must review permissions and contract terms before a pilot with actual patients.
A data-flow map and a list of unresolved access, retention, export and vendor questions.
Define a reference review and a unit of analysis before testing. Separate patients across development and evaluation sets. Use an image-quality flag to study sensitivity, specificity and false negatives. Document disagreement between reviewers and inspect performance by capture condition.
In 40 synthetic images, reviewers flag 10 as unusable. The model flags 8 of those and 3 of the 30 usable images. Build the confusion matrix, calculate sensitivity and specificity, and decide how the two misses should be handled.
An evaluation sheet with denominators, error examples, a proposed threshold and a reason to accept or revise it.
Compare density overlays, region boundaries and trend displays against the original images. Examine the effects of lighting shifts, hair length and inconsistent landmarks. Distinguish a visualization from a measured result and recognize when an output cannot support a clinical conclusion.
Review a mock scalp series with improved density reported after a changed hair part, and a skin series with changed exposure. Choose accept, correct, recapture or escalate and justify each choice.
An annotated image-review form with discrepancy codes and a clinician sign-off field.
Use a structured prompt with explicit source fields and permitted transformations. Require missing facts to stay missing. Compare the draft against the supplied history, exam and plan for invented findings, negation errors, laterality and unsupported recommendations.
A synthetic note says the patient denied itching and no family history was recorded. A draft reverses the symptom and adds a family history. Trace both errors to the source, correct the note, then test the revised prompt on a second case.
A versioned prompt, original draft, corrected note and line-by-line review trail.
Design a review queue and assign responsibility to capture staff, clinicians and supervisors. Define actions for an empty image set, two mismatched patients, an unavailable model, and conflicting clinical findings. Record overrides and stop publication of an unsigned note.
Run a tabletop exercise in which one image belongs to the wrong visit. Follow the record from detection through quarantine, correction, re-review and return to normal work.
An exception playbook, approval checklist and audit entry that reconstructs what happened.
Set a baseline for documentation time, correction rate, recapture rate and unresolved errors. Compare first-pass and signed outputs. Record software versions and establish who can pause the pilot when a material change or recurring error is observed.
Compare two fictional weekly reports: drafting time falls while clinician corrections rise. Decide whether to expand, revise or stop the pilot and identify the additional evidence needed.
A pilot scorecard, change log and release decision with an owner and review date.
A two-provider clinic wants to compare follow-up scalp images and draft visit notes. One pair has inconsistent lighting, one chart lacks an examination finding, and one draft proposes treatment the clinician never documented. Build a workflow that catches all three before sign-off.
Deliver a capture standard, a small evaluation with denominators, a corrected note, an exception log and a pilot plan. Demonstrate a normal visit and one failed case from intake through human approval.
Ask about the planned elective. Admissions will share scheduling and tuition when available.