BRENNEN

Industries / Health / Healthcare / Dental

Prevention and gum care

Prevention and gum care is the work of stopping decay and gum disease before they need drilling or surgery: fluoride in water and toothpaste, keeping teeth clean, and measuring and treating the gums.

6

Milestones achieved, from 1945 to 2025.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1945

    Fluoride added to city water

    In 1945 Grand Rapids, Michigan, became the first city in the world to add fluoride to its drinking water, according to the US National Institute of Dental and Craniofacial Research (NIDCR). The idea came from H. Trendley Dean, head of the dental hygiene unit at the National Institute of Health, whose studies found that up to 1.0 part per million of fluoride in water did not cause enamel fluorosis (mottled, stained teeth) in most people. In a 15-year study of almost 30,000 schoolchildren, decay among children born after fluoridation began fell by more than 60% after 11 years.

    nidcr.nih.gov

    Credit

    H. Trendley Dean and the Grand Rapids City Commission, with the US Public Health Service

  2. 1965

    Plaque alone inflames gums

    Harald Löe, Else Theilade and S. Börglum Jensen had healthy volunteers stop cleaning their teeth. Plaque, the film of bacteria that builds up on teeth, accumulated and the gums became inflamed (gingivitis), and the inflammation went away once cleaning resumed. A 2025 systematic review calls the 21-day version of this "experimental gingivitis" model the gold standard, and it is still widely used to study gum inflammation and to test preventive products and treatments.

    api.crossref.orgpmc.ncbi.nlm.nih.gov

    Credit

    Harald Löe, Else Theilade and S. Börglum Jensen

  3. 1989

    Keep 20 teeth at 80

    Japan's Ministry of Health and Welfare and the national dental association launched the 8020 campaign, which asks people to keep at least 20 of their own teeth at age 80, the number taken as necessary for adequate chewing. The ministry's national dental surveys put the share of 80-year-olds with 20 or more teeth at 15.3% in 1999, 51.2% in 2016 and about 61.5% in 2024; the ministry notes that the 2024 figure was weighted differently from earlier surveys.

    pmc.ncbi.nlm.nih.govmhlw.go.jp

    Credit

    Japan's Ministry of Health and Welfare and the Japanese Dental Association

  4. 2018

    Gum disease reclassified

    The American Academy of Periodontology (AAP) and the European Federation of Periodontology (EFP) published a new classification of periodontal (gum) diseases, agreed by experts from around the world at a workshop in Chicago on 9 to 11 November 2017 after 19 review papers and four consensus reports. It replaced the 1999 classification, gave clinicians case definitions and diagnostic criteria, and added a matching scheme for disease around dental implants.

    ebi.ac.uk

    Credit

    Jack G. Caton, Gary Armitage, Tord Berglundh, Iain L. C. Chapple, Søren Jepsen, Kenneth S. Kornman and the workshop participants

  5. 2021

    Calgary votes fluoride back

    Calgary, Canada, had fluoridated its water since 1991 and stopped on the order of its city council on 19 May 2011. Lindsay McLaren and colleagues then compared Grade 2 children in Calgary with those in still-fluoridated Edmonton: in 2018 and 2019, 64.8% of Calgary's children had decay experience in their baby teeth against 55.1% in Edmonton, and the gap had widened since fluoridation stopped. In the 2021 municipal election 62% of voters chose to bring fluoride back, and it returned on 30 June 2025.

    Lessonthe harm of stopping showed up slowly, in children born after the change, so it took most of a decade to measure and four more years to reverse.

    calgary.caebi.ac.uk

    Credit

    Lindsay McLaren (University of Calgary) and colleagues; Calgary's voters and council

  6. 2025

    Two states ban fluoridation

    Utah's law HB 81, in force from 7 May 2025, bars anyone from adding fluoride to a public water system and lets pharmacists prescribe fluoride instead. Florida also enacted a statewide ban in 2025, according to the oral health group CareQuest, and 15 states had bills to restrict or ban fluoridation by February 2026. On 31 October 2025 the FDA moved to limit the sale of ingestible fluoride prescription products for children under three and for older children at low or moderate risk of decay, and sent notices to four companies.

    le.utah.govcarequest.orgagd.org

    Credit

    the Utah Legislature (sponsors Stephanie Gricius and Kirk A. Cullimore), the Florida Legislature and the FDA

Watch for

  1. Probing depth without a probe

    Probing is the usual gum exam: a thin ruler-like probe is slid between gum and tooth at six points around every tooth to measure the pocket. Minhui Tan, Zhiming Cui and colleagues at the Shanghai PerioImplant Innovation Center and partner engineering schools say their PerioAI system measures the same thing from an intraoral scan and a cone-beam CT scan, by finding the distance from gum edge to bone. They report testing it on 2,507 patients from several centers, an error of 0.040 mm in its digital measurement, and results comparable to clinical probing. A 2026 study of 53 patients by the same group found that adding PerioAI's depths to a panoramic X-ray raised one periodontist's agreement on prognosis with decisions made from a full probing chart from 77.6% to 84.7%.

    Achieved whena group outside the developers, with its own scanners and patients, reports scan-derived depths within 1 mm of probe readings at most sites.

    pmc.ncbi.nlm.nih.govebi.ac.uk

    Claimed by

    Minhui Tan and colleagues, in Cell Reports Medicine, June 2025

  2. EPA review of fluoride limit

    The US Environmental Protection Agency (EPA) is running an expedited review of the health risks of fluoride in drinking water. On 2 August 2026 it released the protocol for its new toxicity assessment, calling it a significant milestone and saying the work is moving "years ahead of schedule"; the next routine review would not have been due until 2030. The assessment is to inform "potential revisions" to the legal limit of 4.0 milligrams per liter, set in 1986 and last reviewed in 2024.

    Achieved whenEPA publishes the draft assessment for public comment and then either proposes a new limit in the Federal Register or states that it will keep the old one.

    epa.gov

    Claimed by

    US EPA, August 2026

Open

  1. 01

    A test for voice charting

    A public set of recordings in which volunteers dictate full periodontal charts (six pocket depths per tooth, plus bleeding, gum recession and loose teeth) while working on a typodont, a plastic model of the teeth in a jaw frame, with real clinic noise from suction and ultrasonic scalers recorded separately and mixed in at set levels. Each recording comes with the chart it should produce, and a script scores any speech-to-chart system by the share of chart cells it gets wrong.

    The brief
    Credit · unclaimed
  2. 02

    A robot probes a typodont

    A desktop robot arm that holds a periodontal probe through a small force sensor and measures pocket depths on a model with known answers: a 3D-printed jaw with pockets of set depths from 1 to 9 mm, covered by a soft silicone gum. The robot pushes at the standard light probing force of about 0.20 to 0.25 newtons, reads depth from its own motion and from a camera, and writes the chart.

    The brief
    Credit · unclaimed

Open 01 · the brief

A test for voice charting

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public set of recordings in which volunteers dictate full periodontal charts (six pocket depths per tooth, plus bleeding, gum recession and loose teeth) while working on a typodont, a plastic model of the teeth in a jaw frame, with real clinic noise from suction and ultrasonic scalers recorded separately and mixed in at set levels. Each recording comes with the chart it should produce, and a script scores any speech-to-chart system by the share of chart cells it gets wrong.

Why it is still open

Bola AI advertises a "99% accuracy rate" for its voice perio product without saying how it was measured. The published tests are small: in Taiwan, Lo and colleagues found a two-step system built on GPT-4o-transcribe got 67.4% of fields right over 15 recordings, and only 53.3% of probing depths; in Korea, Kim and colleagues found complete phrases recognized in 36.7% of cases with Google's engine and 52.5% with Naver's. A 2026 review in the British Dental Journal found only one dental study of speech-to-text accuracy. Open code exists without test data: ToothStream on GitHub (May 2026) sets itself a target of 98% accuracy on pocket depths but publishes no recordings to check that against.

github.combola.aiebi.ac.ukebi.ac.ukebi.ac.uk

Built with what exists

A typodont and a periodontal probe; a phone and a clip-on microphone; speech-to-text engines such as ElevenLabs Scribe v3, which its maker says gets about 12 words in 100 wrong averaged over 102 languages (see the Note "Fewer word errors"), and open models such as Whisper; a language model that turns the transcript into a chart; the commercial voice-charting products used through their normal trial accounts.

  1. First step

    Write ten chart scripts, record three volunteers reading them with and without scaler noise, and score two engines.

  2. Done looks like

    At least 200 full-mouth dictations by at least 20 speakers in at least four languages, each at three noise levels, published with their true charts; cell error rates for at least five systems, including two commercial ones, with probing depths scored separately.

Open 02 · the brief

A robot probes a typodont

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A desktop robot arm that holds a periodontal probe through a small force sensor and measures pocket depths on a model with known answers: a 3D-printed jaw with pockets of set depths from 1 to 9 mm, covered by a soft silicone gum. The robot pushes at the standard light probing force of about 0.20 to 0.25 newtons, reads depth from its own motion and from a camera, and writes the chart.

Why it is still open

3Shape holds a US patent, granted on 4 February 2025, for reading the depth of a hand-held probe from cameras worn on dental loupes, at a probing force of about 20 to 25 grams. A 2026 review of oral examination robots describes robots that probe pockets under force limits of 20 to 25 grams. PerioAI skips the probe altogether. The review concludes that how such robots compare with experienced dentists at tasks including periodontal probing "has yet to be systematically validated", and the two papers it cites for robot probing describe a swab-sampling robot for the mouth and a robotic probe tip with capacitive sensors. A search of journals, patents, GitHub and the web found no published test that scores a robot against a model with known pocket depths.

patents.google.compmc.ncbi.nlm.nih.gov

Built with what exists

A low-cost six-axis arm trained by demonstration with Hugging Face's LeRobot (see the Note "No download") or Runway's open Praxis-1 robot model (see the Note "Open robot model weights"); a miniature load cell; a resin 3D printer and silicone for the gum; a USB camera; pocket depths taken from the CAD file.

  1. First step

    Print one quadrant with 42 sites of known depth, and check by hand with a probe on a kitchen-scale load cell that the silicone gum behaves like gum at 0.25 N.

  2. Done looks like

    On a model with 168 sites (28 teeth, six sites each), the robot's readings are within 1 mm of the true depth at 95% of sites over three full filmed runs, with force never above 0.25 N, and are compared with two hygienists probing the same model.