idea-001 · feasibility deck
Every slide states a condition that would have to be true, then reports where it stands using the candidate file's own claim and its own confidence tag. A deck never upgrades a tag, invents a number, or recommends anything, and there is no ask slide.
The chair: the engineers who have to sense microvolts on the same probe that is delivering stimulation, and ship it as a home device with an app attached.
How to read this: every slide is a condition, not a conclusion. A bracketed claim reference points into knowledge-base/candidates/idea-001.md — the number is the claim's position in its ## Claims list, the tag is copied from it and never adjusted here. A [no claim] marker means nothing in the file speaks to the condition at all. See docs/deck-spec.md.
What changed since the 2026-08-24 build: the research lane added a second InControl clearance whose biofeedback modality is manometry, not EMG [claim 37: verified], a recall scan across both product codes [claim 38: unconfirmed], and the Cochrane evidence base for the mechanism [claim 43: unconfirmed]. Every deck-surfaced engineering assumption went through the Verifier and came back unconfirmed on structural grounds — no connector here reaches bench performance. Rebuilt from the file, not patched.
Would have to be true: One intravaginal probe would have to deliver neuromuscular stimulation and record surface EMG from the same tissue, close the loop between them in real time, and stream adherence and EMG trends to a clinician-facing app.
Where it stands: The function is stated in the candidate's frontmatter and the regulatory container for it is verified: product code KPI, Class II, 21 CFR 876.5320 [claim 1: verified], [claim 2: verified]. No performance specification of any kind appears in the file — no stimulation parameters, no EMG sensitivity, no sampling requirement [no claim] — so this deck states none either.
What would settle it: The cleared predicates' 510(k) summaries for the parameter ranges FDA has already accepted in this class — including K134020, now identified and verified [claim 37: verified]; IEC 60601-2-10 for nerve and muscle stimulator requirements.
If it's false: There is no spec to design or test against, and the pathway claims below describe permission to sell a device nobody has defined.
Would have to be true: Closed-loop EMG-gated stimulation would have to beat the open-loop stimulation already on the market, and beat the exercise programme the patient has already failed, by a margin a clinician would change practice for.
Where it stands: The evidence base arrived since the last build and it does not support the premise. A Cochrane review of 56 randomised trials (n=3,781) found electrical stimulation probably more effective than no active treatment (RR 2.31, 95% CI 1.06-5.02) but low-quality evidence of no difference against pelvic floor muscle training (RR 0.85, 95% CI 0.70-1.03) [claim 43: unconfirmed]. This candidate's users are defined as women who already failed that training, so the pooled comparison runs against the product's own indication. The candidate's differentiating premise remains this deck's write-back, and the Verifier could resolve neither half of it [claim 33: unconfirmed]. What the file verifies is adjacent and weaker: devices in this class are cleared for this indication [claim 4: verified] — clearance being substantial equivalence, not demonstrated superiority. The differentiation argument therefore rests on data and billing workflow [claim 22: unconfirmed] rather than on any mechanism evidence in this file.
What would settle it: A head-to-head of closed-loop EMG-gated against open-loop stimulation in the post-PFMT-failure cohort. Nothing in the Cochrane pool tests the gating mechanism itself, so the existing literature removes the fallback argument without touching the specific claim.
If it's false: The closed loop is engineering cost with no clinical return, and the simplest competitive response — an existing cleared external stimulator [claim 4: verified] — already exists.
Would have to be true: A 510(k) against a KPI predicate would have to be the route, with electrical safety, biocompatibility for mucosal contact, and software documentation as the build burden.
Where it stands: The class and regulation are verified [claim 1: verified], [claim 2: verified]. Three notes, and the predicate one is now better than it was. First, "510(k) rather than De Novo" is unchecked, not confirmed — openFDA publishes no De Novo endpoint and this repo holds no cached listing, so the connector could not look [claim 6: unconfirmed]. Second, the predicate pool grew: the research lane found a second, distinct InControl clearance, K134020, whose Indications for Use name male and female urinary and fecal incontinence with visual biofeedback via manometry [claim 37: verified] — materially different framing from K133826, whose usability as a predicate was downgraded once its cleared indications could actually be read [claim 3: unconfirmed]. Third, the file contains a clean lesson in indication scope: a separate clearance was required to go from stress to urge incontinence [claim 5: verified], so any widening is a second submission, priced accordingly.
What would settle it: The De Novo listing cached in data/; the 510(k) summaries for K134020 and the ELITONE clearances for the test batteries accepted; FDA's software and cybersecurity guidance for the app's documentation level.
If it's false: A De Novo path changes cost and calendar by a multiple, and the capital assumption [claim 30: unconfirmed] is wrong at its foundation.
Would have to be true: Microvolt-scale EMG would have to be recoverable on the same probe that is delivering stimulation — with artifact rejection good enough to gate stimulation on the signal in real time.
Where it stands: Unbacked, and the Verifier's disposition is that it is structurally unbackable here: a bench/engineering feasibility claim that "no connector in this repo reaches" in any of its shapes [claim 34: unconfirmed]. The stimulus is orders of magnitude larger than the signal, at the same electrode-tissue interface, and "concurrent" is the candidate's own word. The most useful new fact on this deck bears directly on it: the incumbent's own second clearance delivers biofeedback by manometry — pressure sensing — rather than EMG [claim 37: verified]. That is not evidence the EMG approach fails, and this deck does not read it as such; it is evidence that a cleared competitor in exactly this class chose a different sensing modality for the same job, which is worth understanding before committing to the harder one. Everything distinctive depends on solving it: the biofeedback the user is meant to see [claim 33: unconfirmed], the longitudinal dataset that carries the moat [claim 22: unconfirmed], and — if PDAC guidance reads as the file claims — the integrated monitoring that qualifies the device for its code at all [claim 8: unconfirmed], [claim 45: unconfirmed].
What would settle it: A benchtop rig on tissue phantom: stimulation artifact amplitude at the sense electrodes, blanking and recovery time, recoverable EMG bandwidth between pulses, and whether the residual signal supports closed-loop gating at the intended stimulation duty cycle. Alongside it, a read of K134020's summary for why manometry was chosen.
If it's false: The device degrades to an open-loop stimulator with an adherence log — which is a cleared, commodity product in a class taking roughly 14 new clearances since 2024 [claim 25: unconfirmed] — or it becomes a manometry device, which is the incumbent's product [claim 37: verified].
Would have to be true: Artifact rejection, current-density limits at the mucosal interface, electrical safety and data-capture reliability would all have to be demonstrated on the bench before any patient uses this.
Where it stands: No bench claim exists in the file [no claim]. The two conditions this deck put on record cover the sensing problem [claim 34: unconfirmed] and the materials and cleaning regimen [claim 36: unconfirmed], and the Verifier returned both as structurally unreachable — no connector performs or reaches biocompatibility test results. The clinical premise a study would eventually have to test is also unbacked [claim 33: unconfirmed], which sets the order of work: bench first, and the clinical question is not this repo's to greenlight. One adjacent signal is now on record and should not be over-read: a recall scan across both product codes KPI and QAJ from 2010 to 2026 found no recall record [claim 38: unconfirmed], which is a clean class-safety history rather than evidence about this device.
What would settle it: IEC 60601-2-10 and ISO 10993 define most of the battery; the predicates' summaries show what FDA accepted in this class.
If it's false: Any move to human use is premature. The boundary is explicit: what happens after bench evidence is a Tier 3 decision, and no agent here makes it.
Would have to be true: A single-patient intravaginal device would have to be cleanable at home, materially safe for repeated mucosal contact, and paired with an app that captures data reliably enough to bill on.
Where it stands: Materials and cleaning are on record and structurally uncheckable here [claim 36: unconfirmed]. The data-capture requirement is the interesting one, because it is an engineering requirement created by a billing rule: the RTM device-supply code is claimed to require 16 days of data in 30 [claim 13: unconfirmed], so the app has to hit that threshold automatically, on a home user's phone, without the clinician chasing it [claim 35: unconfirmed]. And that requirement is unverifiable in this repo by construction — the code descriptor is AMA-licensed [claim 13: unconfirmed] — so engineering would be building to a spec the pipeline cannot read. The related coding question did move: the sourced PDAC guidance says E0740 is for stimulators with integrated monitoring and that monitoring-without-stimulation devices were not intended to be included [claim 45: unconfirmed], which points the same way as the file's own reading [claim 8: unconfirmed].
What would settle it: ISO 10993 biocompatibility for the wetted materials; a licensed AMA CPT source for the data threshold; a connectivity and drop-out analysis against the real user population.
If it's false: The monitoring revenue line fails on data completeness rather than on clinical grounds — an unglamorous way to lose the incentive that makes prescribers prescribe [claim 24: unconfirmed].
Would have to be true: Closed-loop EMG-gated intravaginal NMES would have to be buildable without infringing live art, with the 1990s-era core stimulation patents genuinely expired.
Where it stands: Unresolved, and the reason is a tooling limit rather than an unknown fact: no connector call was possible and none was made — the patent connector takes a specific patent number and cannot run a landscape search, and this claim names no number to pass it [claim 21: unconfirmed]. So this is not "we looked and found nothing", it is "the question cannot currently be asked in this repo". The likely blocking surface is the novel element — EMG-gated closed-loop control — not the commodity stimulator the moat claim already concedes is undifferentiated [claim 22: unconfirmed].
What would settle it: A patent landscape search on EMG-gated pelvic floor stimulation and on intravaginal probe geometries, by hand or with a tool this repo does not have; expiry verification on the named 1990s families; a Tier 3 legal opinion to settle it.
If it's false: Automatic kill under the Stage 5 gate — the only condition on any of these three decks with that property, and the one this repo is currently least able to check.
Every condition above with nothing verified behind it.
data/.
The pattern the last build reported has one crack in it now. The verified claims in this file are still overwhelmingly about permission and payment — [claim 1: verified], [claim 2: verified], [claim 4: verified], [claim 5: verified], [claim 7: verified], [claim 10: verified] — but one verified claim is finally about a device doing something: K134020's manometry biofeedback [claim 37: verified]. It is a competitor's engineering, not this candidate's. Every engineering condition specific to this device is still unbacked, and the Verifier has now confirmed that is structural rather than an unspent connector call.
If only one thing from this chair could be checked: whether microvolt EMG can be recovered on the same probe that is stimulating, well enough to gate the stimulation in real time [claim 34: unconfirmed].
It is the one place where this candidate is more than a commodity stimulator, and the week's new evidence sharpened rather than moved it: the mechanism's clinical fallback weakened [claim 43: unconfirmed], and the cleared incumbent turns out to deliver biofeedback by manometry instead [claim 37: verified]. The clinical premise [claim 33: unconfirmed], the dataset moat [claim 22: unconfirmed], and possibly the code's integrated-monitoring requirement [claim 8: unconfirmed] all rest on it. It is also cheap and fast to settle — a bench rig on a phantom, no patients, no regulator, no payer — and it can be answered before a dollar of the capital assumption [claim 30: unconfirmed] is spent.
Naming it is not a recommendation, a gate, or a kill.