Nonunion PEMF Bone Growth Stimulator — feasibility

The chair: engineering. The people who have to build it, test it, manufacture it, and produce the evidence a reviewer will ask for.

How to read this: every slide is a condition, not a conclusion. A bracketed claim reference points into knowledge-base/candidates/idea-018.md: the number is the claim's position in its ## Claims list and the tag is the one the candidate file carries today, copied and never adjusted here. A no-claim marker means nothing in the file speaks to the condition. See docs/deck-spec.md.

What changed since the last build. The 2026-09-29 Verifier pass read the special controls at 21 CFR (Title 21 of the Code of Federal Regulations) 890.5870(b) in full, and they are a build spec. Two things in them reshape this deck. (b)(1) requires clinical data with imaging demonstrating fusion, so no 510(k) (premarket notification) under this regulation is bench-only. (b)(6) requires labeling comprehension testing, a human-factors study for a device used at home. The deck-surfaced engineering claims from the last build (38–41) moved from unverified to unconfirmed: a Verifier looked and no connector can settle them.

Slide 1 — What the device actually has to do

Would have to be true: A battery-powered, garment-integrated coil thin enough to wear under ordinary clothing would have to deliver a pulsed electromagnetic field across the fracture gap of an adult long bone. Its field magnitude at the gap would have to match the incumbents'. It would have to run for the prescribed hours per day, on a single charge, over a multi-month course. It would also have to log actual field-on minutes in a form that cannot be falsified without detection. The field magnitude, the hours per day and the course length are blanks: no claim carries any of them.

Where it stands: The indication the spec serves sits inside the regulation's own language claim 2verified. That the thin coil reaches the incumbents' field magnitude within the (b)(2) limits is unchecked and needs bench data claim 39unconfirmed. The Verifier recorded that (b)(2) names categories of testing but sets no numeric field or thermal limit, so there is no regulatory number to design to. A full day's treatment on one charge is a condition this deck put on record claim 49unverified. Tamper-evident logging is also on record here claim 48unverified. The incumbents' field parameters and treatment schedules are not in the file [no claim].

What would settle it: Retrieve the incumbents' labelling or the summaries of their PMAs to get the field parameters and the prescribed daily treatment time. No connector here reads PMA (premarket approval) documents, so a human would have to. Then take a bench measurement of the prototype coil's field at a depth representative of a long-bone fracture gap.

If it's false: Without the incumbents' parameters there is no target for Slides 4 and 5. If the thin coil cannot reach the target, the low-profile differentiator and the equivalence argument on Slide 3 fail together.

Slide 2 — The mechanism that has to hold

Would have to be true: Pulsed electromagnetic field stimulation, at the magnitude and schedule in Slide 1, would have to improve union rates in established long-bone nonunion relative to standard fracture management alone, in controlled evidence. The size of the effect it has to reach is a blank.

Where it stands: Unread. The condition that controlled evidence shows PEMF (pulsed electromagnetic field) improves union in established nonunion is unchecked claim 40unconfirmed. The Generator retrieved nothing claim 25unconfirmed. The Verifier's sweep showed the literature exists: 80 matches on PEMF and nonunion, including a 2024 systematic review update. None of it has been read, and a title is not a finding. The only appraised trial in the file is negative and adjacent. TRUST (PMID (PubMed identifier) 27797787; 501 patients, sham-controlled) found LIPUS did not accelerate radiographic healing of operatively fixed tibial fractures claim 24verified. That is fresh fracture and ultrasound, not this mechanism in this indication. The long market history of the device class supports plausibility and does not show effect size claim 5verified.

What would settle it: A literature-detail read of the leads recorded under claim 25unconfirmed, starting with PMID 39387850, to find which of them address established long-bone nonunion as distinct from fresh fracture. Each result that holds up goes into its own claim line for the Verifier.

If it's false: If PEMF shows no controlled benefit in established nonunion, Factor 6 has nothing to stand on. The (b)(1) clinical data on Slide 3 would have to be produced from scratch, and no reliance on an incumbent's data would carry a result the literature does not show.

Slide 3 — What the pathway forces you to build

Would have to be true: The device would have to be built and evidenced to all seven special controls in 21 CFR 890.5870(b). It would have to reach 510(k) clearance with a predicate FDA (US Food and Drug Administration) accepts. Its clinical-data requirement would have to be met either by relying on an incumbent's data or by a study small enough to fund. A nonunion outcome would have to count as the "fusion" the imaging control names.

Where it stands:

• The controls, read verbatim claim 3verified. (1) Clinical data, with imaging demonstrating fusion at the treatment site. (2) Non-clinical testing of output delivery, thermal safety, signal characteristics and use-life reliability. (3) Biocompatibility of patient-contacting parts. (4) Electrical safety and EMC. (5) Software verification, validation and hazard analysis. (6) Labeling comprehension testing. (7) Warnings and a clinical-data summary in the labelling.
• The fusion wording: the imaging control is written for fusion, and this candidate's indication is nonunion. Whether FDA reads radiographic union of a nonunion as satisfying it is a condition this deck put on record claim 46unverified.
• Clinical-data reliance: the final order says an applicant "may be able to rely on clinical data from a legally marketed non-invasive bone growth stimulator with the same intended use". The Verifier quotes this under claim 38unconfirmed. Whether a new entrant with a different coil can in fact meet (b)(1) that way, without running its own study, is a condition this deck put on record claim 47unverified.
• The predicate: no 510(k) has ever been cleared under LOF or LPQ claim 20verified. Whether a pre-reclassification PMA device can serve as the predicate is unresolved claim 38unconfirmed.
• The scope: whether the narrow label is what keeps the device in 510(k) territory at all is an FDA judgement claim 7unconfirmed.

What would settle it: One FDA pre-submission (Q-Sub) that asks the predicate, reliance, fusion-wording and scope questions together. That is a human act, and claims 7 and 38 both name it.

If it's false: If reliance is refused, the device needs its own clinical study with imaging endpoints. That changes the build from a hardware-and-bench programme to a clinical programme, which is Factor 7's damage and the viability deck's Slide 4. If there is no predicate as well, the "cheap pathway" premise of the candidate collapses into De Novo (De Novo classification request) territory.

Slide 4 — The hardest unknown

Would have to be true: The garment-integrated coil would have to be thin and flexible enough to wear under clothing and still deliver, at the depth of an adult long-bone fracture gap, the field magnitude the incumbents' rigid coils deliver. It would have to do so within (b)(2)'s thermal-safety and signal-limit testing, with the coil in contact with the patient's skin for hours a day.

Where it stands: Unchecked, and no connector can check it; it needs bench data claim 39unconfirmed. The incumbents' field magnitude, the number the coil has to hit, is not in the file [no claim]. The (b)(7) labelling control requires warnings against use on compromised skin and for patients with implanted devices claim 3verified. How often the target patients carry fixation hardware at the nonunion site, and how the coil's field and heating interact with it, are both absent from the file [no claim].

What would settle it: A bench rig: prototype coil over a tissue phantom, with field magnitude mapped against depth and thermal rise logged over a full treatment session. Run it against the incumbent target from Slide 1 once a human retrieves that target. For implanted hardware, a field-and-heating test with representative fixation metal in the phantom.

If it's false: If the thin coil cannot reach the field, the product is either a conventional coil, with no form-factor differentiator, or a device claiming equivalence it cannot show on the bench. Slide 3's reliance argument goes with it.

Slide 5 — Bench evidence before anything lives

Would have to be true: Before any patient is involved, a bench programme would have to show four things. The coil meets the Slide 4 field target within thermal limits. A full treatment day runs on one charge. Output and reliability hold over the course length. The adherence log cannot be falsified without detection, whether by edited entries, by wear with the field off, or by a manipulated clock.

Where it stands: (b)(2) makes output delivery, thermal safety, signal characteristics and use-life reliability mandatory test categories claim 3verified. Whether the design passes them is unchecked claim 39unconfirmed. Single-charge endurance is a condition this deck put on record claim 49unverified. So is detectable falsification of the log claim 48unverified. That condition matters twice over. The log is the feature whose commercial value is an explicit assumption claim 30unconfirmed, and a log that can be gamed has no documentation value to anyone.

What would settle it: A bench test plan, drafted by a human engineer and mapped line-by-line to (b)(2)(i)–(iv) and (b)(5). For the log, an adversarial test in which testers try each falsification route and check whether the exported report shows it. Tier 3 sits directly downstream of this slide, and nothing here greenlights a clinical or animal study.

If it's false: If the log can be defeated quietly, the differentiator the viability deck's Slide 5 leans on has no value. If endurance fails, the garment form factor and the daily schedule are in conflict and one of them has to give.

Slide 6 — Making it, and using it

Would have to be true: An adult, alone at home, would have to be able to don, operate and charge the garment coil correctly from the instructions for use, well enough to pass (b)(6) labeling comprehension testing. The coil and its logging electronics would have to be manufactured at a unit cost below what the device is paid.

Where it stands: (b)(6) requires labeling comprehension testing showing the patient can use the device from the IFU (instructions for use) alone claim 3verified. Whether this design passes is a condition this deck put on record claim 50unverified. The payment ceiling is the E0747 (a HCPCS code for durable medical equipment) new-purchase allowance of $4,743.80–$5,580.94 non-rural claim 16verified. That is Medicare's allowed amount, not what the manufacturer realises. Whether a Class II (FDA device class II) device bills on those KF lines at all is carried on the viability deck claim 45unverified. The unit cost has no source claim 41unconfirmed. (b)(3) biocompatibility applies to every patient-contacting textile and component in the garment claim 3verified. No claim addresses whether the garment can be cleaned or laundered, or how long it lasts [no claim].

What would settle it: A formative human-factors study with nonunion patients on a prototype, run before the summative (b)(6) test. A costed bill of materials from a contract manufacturer quote, which is a human act. A wash-and-wear endurance test on the garment integration.

If it's false: If patients cannot use it from the IFU, (b)(6) fails and the 510(k) does not clear, whatever the coil does. If unit cost exceeds realised price, the viability deck's Slide 3 is negative per unit before share is even considered.

Slide 7 — What FTO forbids

Would have to be true: No live US patent would have to cover the PEMF waveform, the garment-integrated coil geometry or the tamper-evident compliance log in a way the design cannot work around. Any workaround would have to be the design, not a degraded version of it.

Where it stands: Unexamined. No patent search of any kind was run claim 26unconfirmed. The belief that the 1970s–80s foundational PEMF patents have expired is an inference about a class of patents, and it says nothing about the two modern differentiators, where live art is most likely claim 27unconfirmed. Unexamined is not clear. The Stage 5 FTO (freedom to operate) condition cannot fire on this record in either direction.

What would settle it: A PatentsView or EPO (European Patent Office) OPS search against Orthofix, Zimmer Biomet, Bioventus and Enovis as assignees, plus a CPC search on wearable-coil and treatment-compliance logging. The patent connector can then check the dates of any number that turns up. An FTO opinion itself is counsel's work.

If it's false: A live patent on garment-integrated coils or compliance logging would remove a differentiator or force a redesign of it. Slides 4 and 5 would be re-specified around the workaround.

Slide 8 — Where this deck outruns the file

Every condition above with nothing verified behind it. Read this slide first. The verified material on this deck is the regulatory text (claims 2, 3, 5, 20), one adjacent negative trial (claim 24verified) and the price (claim 16verified). It says what must be built. Nothing verified says the build is possible.

• Thin coil reaches the incumbents' field within (b)(2) limits — claim 39unconfirmed — bench rig, Slide 4.
• The incumbents' field parameters and treatment schedule — [no claim] — a human retrieves labelling or PMA summaries.
• Full treatment day on one charge — claim 49unverified — bench endurance test (written back by this deck).
• Falsification of the log is detectable — claim 48unverified — adversarial bench test (written back by this deck).
• PEMF improves union in established nonunion — claim 40unconfirmed — literature-detail on the claim-25 leads.
• The nonunion evidence base — claim 25unconfirmed — same sweep.
• Radiographic union counts as (b)(1) "fusion" — claim 46unverified — FDA pre-submission (written back by this deck).
• Clinical-data reliance on an incumbent — claim 47unverified — FDA pre-submission (written back by this deck).
• A PMA device is a usable predicate — claim 38unconfirmed — FDA pre-submission.
• The narrow label keeps the device in 510(k) — claim 7unconfirmed — FDA pre-submission.
• Coil interaction with implanted fixation hardware — [no claim] — bench field-and-heating test.
• Home users pass (b)(6) from the IFU alone — claim 50unverified — formative, then summative, human-factors study (written back by this deck).
• Unit cost below the allowance — claim 41unconfirmed — contract-manufacturer quote.
• A Class II device bills on the KF lines — claim 45unverified — see the viability deck.
• Garment cleaning and durability — [no claim] — wash-and-wear endurance test.
• The log has commercial value — claim 30unconfirmed — see the viability deck.
• FTO — claim 26unconfirmed and claim 27unconfirmed — assignee and CPC search, then counsel.

Slide 9 — The load-bearing condition

If only one thing from this chair could be checked: whether a new entrant can meet the 21 CFR 890.5870(b)(1) clinical-data special control by relying on an incumbent's clinical data for the same intended use, without running its own clinical study claim 47unverified.

This one rather than the others: it decides what kind of programme engineering is building. If reliance holds, the build is hardware, bench testing and a human-factors study. If it does not, it is all of that plus a clinical study with imaging endpoints. The same pre-submission meeting that answers it also answers the predicate (claim 38unconfirmed), fusion-wording (claim 46unverified) and scope (claim 7unconfirmed) questions, so it is the single human act that moves the most of this deck. Reliance would itself depend on the thin coil being close enough to the incumbents' device, so Slide 4's bench rig is the natural next question, not a rival first one.

Naming it is not a recommendation, a gate, or a kill.