orthodontic.ai

Signal log · newest first

Signal log

Signal log: short, dated briefs. Each one links the public page we opened on Oct 31, 2026 (ET). The source summary uses the source's facts; our read is editorial and hedged.

  1. FDA 510(k)

    FDA clears an AI-enabled prediction service inside a clear-aligner treatment-planning system

    What the source says

    K262196 was decided Substantially Equivalent on Sep 25, 2026. The 510(k) summary describes an optional AI-enabled prediction service added to the backend 3D planning software to optimize aligner activation design. Plans are reviewed and approved by the doctor before manufacturing, and the AI model is described as static and non-adaptive.

    Our read

    This is the clearest regulatory marker yet of AI inside a mainstream aligner planning pipeline. The public summary says acceptance criteria were met but gives no performance figures.

    U.S. FDA: 510(k) Summary K262196 (PDF) Related case

  2. Professional guidance

    AAO publishes a position paper on responsible clinical use of AI

    What the source says

    The American Association of Orthodontists announced a position paper from its Task Force on AI, describing the orthodontist as the "Human-in-Command" at the point of care. The AAO says the paper is not regulatory in nature and sets out six core principles.

    Our read

    Useful as a checklist for buyers: who reviews outputs, how vendors validate and monitor models, and how patients are told. It does not rate products.

    American Association of Orthodontists: AAO Publishes Position Paper to Guide Clinical Use of AI

  3. FDA 510(k)

    A second AI cephalometric tool clears 510(k), with mean radial error thresholds stated

    What the source says

    K243005, an AI cephalometric analysis software, was decided Substantially Equivalent on May 30, 2025. Its summary reports passing acceptance criteria of mean radial error at or below 1.5 mm on lateral and 2.5 mm on frontal cephalograms. No clinical testing was required.

    Our read

    A stated error threshold is more useful to a buyer than a percentage alone. It is still the sponsor's own bench study.

    U.S. FDA: 510(k) Summary K243005 (PDF) Related case

  4. FDA De Novo

    FDA creates a new "dental image analyzer" device type for remote monitoring

    What the source says

    FDA granted De Novo DEN230035 on May 17, 2024 for a remote-monitoring app, creating regulation 21 CFR 872.1770 (product code SBC). The decision summary describes prescription home use, with results intended as an aid, not for stand-alone clinical decisions.

    Our read

    A new classification gives later remote-monitoring tools a predicate to follow. The summary also lists limits worth reading before purchase, such as no results on lingual surfaces.

    U.S. FDA: De Novo decision summary DEN230035 (PDF) Related case

  5. FDA 510(k)

    AI cephalometric analysis software cleared; 99 landmarks generated per image

    What the source says

    K231396, an AI cephalometric analysis software, was decided Substantially Equivalent on Jan 31, 2024. The summary says the software generates 99 landmark points per image. In the sponsor's study of 21 landmarks against three orthodontic specialists, with up to 2.0 mm counted as a pass, it reports that 99% of landmarks met the endpoint. Indicated for patients 14 and older.

    Our read

    The pass margin and the reference panel are stated, which helps. The test set size and its source are not in the public summary.

    U.S. FDA: 510(k) Summary K231396 (PDF) Related case