BRENNEN

Industries / Health / Healthcare

Medical spas

Medical spas are clinics, often run like day spas, where doctors, nurses and other licensed staff give cosmetic treatments without surgery, such as wrinkle injections, fillers, laser work and chemical peels.

1998

Margaret Mitchell, a medical office administrator, and Patrick Bitter Sr., a dermatologist, opened Epi Center MedSpa in San Francisco, a dermatology practice run like a spa; by the clinic's own account it was the first, and its founders named the idea a "MedSpa".

epicentermedspa.com

49

Milestones achieved across 7 domains.

12

To watch for: announced, not yet confirmed outside.

16

Open deliverables. Nobody has shipped them.

In one minute

Where it started

In 1998 Margaret Mitchell and the dermatologist Patrick Bitter Sr. opened Epi Center MedSpa in San Francisco and called their mix of a skin clinic and a spa a "MedSpa"; the claim to be first, and to the word, is theirs. A Medical Spa Association followed in 2002. The American Med Spa Association (AmSpa) dates the industry's real start to the US Food and Drug Administration's approval of Botox for cosmetic use that same year.

Where it stands

AmSpa counts 10,488 US medical spas in 2023, up from 8,899 in 2022, with average revenue of $1,398,833 each, and by its count 67% of single-owner practices belong to people who are not physicians, such as nurse practitioners and nurses. Worldwide, the plastic surgeons who answered the International Society of Aesthetic Plastic Surgery's survey reported 17.3 million non-surgical procedures in 2025, led by 7.3 million botulinum toxin injections, with the US, Brazil and China the largest countries by total procedures. South Korea's health ministry counted 2.01 million foreign patients in 2025, and dermatology drew 62.9% of them by the ministry's count.

What is open

Most products and devices are tested by their makers before sale, and few outside checks follow: what is in a bottle of peel or exosome serum, whether goggles block what their labels say, whether clinics claim more than a device was cleared for, and where complications happen. The open jobs need no patient: laboratory and bench tests of products, registers built from published cases and adverse-event reports, and checks of the rules and of who supervises. AI models that read documents in bulk at low cost now let a small team turn thousands of adverse-event reports, clearance summaries, consent forms or clinic web pages into open tables.

Domains

AchievedWatch forOpen

Advances to build with

3 research notes

Open in medical spas

3 of 16 open deliverables

Open · Consultation

Previews checked against results

A public test of the AI tools that show patients a predicted result before injections. Take before-and-after photo pairs that clinics have already taken in routine care, with each patient's written consent for research and an ethics approval, and note the product, amount and area treated. Feed each "before" photo and the treatment record into each preview tool, then compare the preview with the real "after" photo. No one is treated for the study.

Built with what exists

Commercial preview tools such as EntityMed, Perfect Corp's aesthetic simulation and Zenoti's visualizer; Google's open MediaPipe Face Landmarker to measure distances such as lip height and cheek projection; a blinded rating form in which injectors pick the real result out of the preview, the real result and another patient's result.

Credit · unclaimed

Open · Injections

Ask the label, twenty times

A public set of about 300 questions whose answers are printed in the official labels of the toxins and fillers sold in the US and Europe: units per vial, dilution, storage, approved areas and doses, who should not be treated, and whether units convert between brands (they do not). Each AI model that clinics might use as an assistant is asked every question 20 times, and a question counts as passed only if all 20 answers are right. Nothing is injected and no drug is handled.

Built with what exists

Label PDFs from the FDA's Drugs@FDA and device approval databases and the EU summaries of product characteristics; a Python harness that sends each question and checks the answer against the label line; open-weight models such as Mistral Large 4 run on the tester's own machine (Notes: "A trillion open parameters") alongside hosted models; the pass-on-all-20 bar used in Microsoft's ThinkingBox research (Notes: "All twenty times").

Credit · unclaimed

Open · Lasers and light

Eyewear tested past 920 nm

A public table of how much light the goggles and patient eye shields sold for aesthetic lasers and IPL really block, from visible light out to 1,600 nanometers, next to what their labels claim. Blocking is given as optical density, a scale on which each step stops ten times more light. The measurements use a laboratory spectrophotometer and its lamp; no laser or IPL is fired and nobody wears the eyewear.

Built with what exists

A UV-visible-near-infrared spectrophotometer with a near-infrared detector that reads past 1,100 nanometers, of the kind university chemistry and physics departments keep; calibrated neutral-density filters to show how dark a sample the instrument can still read; goggles and eye shields bought from clinic suppliers and online marketplaces; Python for the table.

Credit · unclaimed

Where it sits

Broader

Healthcare