The argument · idea-001

Home Stress-Incontinence Stimulator

The device is indicated for the treatment of stress urinary incontinence in cognitively intact adult women in the home setting, by prescription, following a documented failed four-week trial of pelvic muscle exercise training.

not converged shape: P1 opened 2026-09-13 updated 2026-09-13

Where the argument stands

3/7

chairs have answered
1 abstains · 1 blocks · 1 conditional

live register rows resting on a
refuted or unconfirmed claim

impact reviews
5 commission · 1 pivot

pivots — the shape is at
P1

open requests between
departments

Chairs disagreeing in writing is the output here, not a problem to be resolved. A blocks verdict opens an impact review; it does not kill anything, and no department holds a veto. Killing is Stage 5 arithmetic over verified claims. converged is not an approval either — the only approval in this repo is a human's, at Stage 6.

Where one failure would move several chairs

Drag the diagram sideways to pan →

Only the shared dependencies. Left: register rows that two or more answered chairs are standing on, coloured by the tag the candidate file gives that claim. Right: those chairs, coloured by verdict. All 2 of them are drawn. Rows a single department depends on are a local risk and are listed in full under the register below; these are the ones where one refutation moves several chairs at once.

Mermaid source for this diagram

flowchart TD
    A3["A3 · claim 9<br><small>refuted</small>"]
    A1["A1 · claim 2<br><small>verified</small>"]
    d_regulatory["Regulatory<br><small>conditional</small>"]
    d_market-access["Market Access<br><small>blocks</small>"]
    d_legal-ip["Legal & IP<br><small>abstains</small>"]
    A3 --> d_regulatory
    A3 --> d_market-access
    A1 --> d_regulatory
    A1 --> d_legal-ip
    class A3,d_market-access red;
    class A1 green;
    class d_regulatory t2;

The round table

Each chair answers cold, from its own field, without reading the others' position files. What it states is a verdict, the conditions that would change it, and the requests it raises. A condition bound to a claim is a load-bearing statement, which is why it shows up in the register below.

Regulatory

conditional

backs it only if its conditions hold

[C1] A device with this intended use is eligible for 510(k) clearance and requires neither a De Novo nor a PMA claim 6: unconfirmed
[C2] The closed-loop EMG gating and the RTM data path do not, by themselves, push this device outside the KPI predicate pool into a new intended-use claim unbound

stated 2026-09-13 against P1

Clinical & Scientific

unstated

has not been convened, or has not answered

This chair has not been convened on this opportunity, or has not answered yet. That is the state, not a criticism of it.

no position on file

Market Access

blocks

opens an impact review — not a veto

[C1] NCD 230.8, or a successor national or local policy, names E0740 or an equivalent code as covered for this indication claim 9: refuted
[C2] Pelvic floor muscle status qualifies as "musculoskeletal system status" for RTM billing, so the 98977/98980/98981 path is available as a second revenue leg rather than an assumption claim 14: unconfirmed
[C3] A commercially-insured route exists that does not depend on the Medicare policy at all claim 32: unconfirmed

stated 2026-09-13 against P1

Technology & R&D

unstated

has not been convened, or has not answered

This chair has not been convened on this opportunity, or has not answered yet. That is the state, not a criticism of it.

no position on file

Operations & NPI

unstated

has not been convened, or has not answered

This chair has not been convened on this opportunity, or has not answered yet. That is the state, not a criticism of it.

no position on file

Legal & IP

abstains

will not hold a position on the evidence available

[C1] No live, enforceable US patent reads on closed-loop EMG-gated intravaginal NMES for urinary incontinence in a way that blocks this design claim 21: unconfirmed
[C2] The EMG-plus-adherence dataset is defensible against a fast follower for long enough to matter unbound

stated 2026-09-13 against P1

Commercial

unstated

has not been convened, or has not answered

This chair has not been convened on this opportunity, or has not answered yet. That is the state, not a criticism of it.

no position on file

What this opportunity rests on

The assumption register: the subset of claims that are load-bearing, each with who depends on it and what breaks if it is false. Rows arrive two ways — the front end writes one, or a chair binds a condition to a claim, which is the definition of load-bearing.

A1 claim 2: verified

If it is false: the device is not a 876.5320 Class II device and the whole pathway argument restarts on a different regulation

Depends on it: Regulatory, Legal & IP

A2 claim 6: unconfirmed

If it is false: no 510(k) route for this device; a De Novo adds 12-18 months and a new evidence burden before any revenue

Depends on it: Regulatory

A3 claim 9: refuted

If it is false: the coverage argument stops being a restatement of existing national policy and becomes a new one — and the narrow indication was bought specifically to purchase that restatement

Depends on it: Market Access, Commercial, Regulatory

A4 claim 16: refuted superseded by p1

If it is false: the CPT 90912 volume leg of the sizing chain is 28,093, not the 200,000 assumed — the chain's base is an order of magnitude out

Depends on it: Commercial

A5 claim 17: refuted superseded by p1

If it is false: the E0740 volume leg is 4,130, not 10,000 — same chain, same base

Depends on it: Commercial

A6 claim 21: unconfirmed

If it is false: if a live patent reads on closed-loop EMG-gated intravaginal NMES, the mechanism that differentiates this device is the mechanism that blocks it

Depends on it: Legal & IP

A7 claim 32: unconfirmed

If it is false: P1's entire premise fails: if the addressable population is Medicare-primary after all, routing around Medicare routes around the market

Depends on it: Commercial, Market Access

A8 claim 14: unconfirmed

If it is false: the RTM billing path closes and the recurring-revenue leg of the money model goes with it, leaving a one-time device sale

Depends on it: Market Access

What failed, and what it cost

When a registered statement comes back refuted or unconfirmed, an impact review fires. It has exactly three available outcomes and check_loop.py refuses a record that invents a fourth.

IR-1 claim 9 came back refuted — 2026-09-13 commission

The statement: [claim 9: refuted] — that Medicare NCD 230.8 covers pelvic floor electrical stimulation with a non-implantable stimulator for stress and/or urge urinary incontinence in cognitively intact patients who failed a documented four-week trial of pelvic muscle exercise training.
What the check found: the local CMS Medicare Coverage Database cache, now populated at 646 rows, contains no policy naming E0740. The tag moved from unconfirmed (missing-cache) to refuted on 2026-08-30 when the cache landed. The Verifier's own note says to read the caveat before treating it as a clean kill, and the caveat is the point: the check was a code lookup, and NCD 230.8 is a real national coverage determination that may simply not reach us through E0740.
Who depended on it: register row A3 — Market Access, Commercial, and Regulatory. The regulatory dependency is the non-obvious one and the expensive one: the intended use was drafted to mirror NCD 230.8's eligibility language, which is documented in the candidate's own indication-scope note.
What it costs: the narrow indication is currently buying nothing. It was traded for a coverage argument that would be a restatement of existing national policy; if the policy does not reach this code, the indication has been narrowed — shrinking the population in [claim 19: unconfirmed] — in exchange for an evidence saving alone, and Regulatory says there is no further class benefit in narrowing because 876.5320 is already the floor.
Outcome: commission
Because: the distinction between unconfirmed and refuted is doing real work here and it is worth one retrieval to resolve. The refutation is a statement about a cache and a code binding, not about whether NCD 230.8 exists or what it says. Commissioning against a refuted claim is usually re-litigating a finding, and this is the exception the protocol's asymmetry allows for: the finding is about our lookup, not about the world. REQ-0001 is that commission. If the full text comes back and does not reach this indication, this review reopens and the honest outcome becomes pivot, not absorb.

IR-2 claims 16 and 17 came back refuted — 2026-09-13 pivot

The statement: [claim 16: refuted] and [claim 17: refuted] — the two Medicare volume legs at the base of the sizing chain.
What the check found: against the CMS utilization cache for 2024, CPT 90912 ran 28,093 allowed services against a claimed 200,000 (overstated by 612%), and E0740 ran 4,130 allowed units against a claimed 10,000 (overstated by 142%). Both outside the 25% tolerance. Both are Medicare fee-for-service only, which the Verifier flags as a **floor on US volume, not a market size** — the figures exclude Medicare Advantage, commercial, Medicaid, VA and cash-pay.
Who depended on it: register rows A4 and A5 — Commercial, and through the market model every number downstream of it.
What it costs: the sizing chain's base is an order of magnitude out, so every band built on it is wrong by that factor. But the finding is narrower than it looks and the narrowness is the whole opportunity: what was refuted is a claim about *Medicare fee-for-service volume*, and what was assumed was that Medicare FFS volume was the right base to size from.
Outcome: pivot
Because: the refutation is a real finding about the world and commissioning more work against it would be re-litigating it — 28,093 is 28,093. What is wrong is not the number but which population the opportunity was resting on. Market Access had independently reached the same door from the coverage side: if the national policy does not hold and the Medicare volumes are a floor rather than a market, both chairs are pointing at the commercially-insured population. That is P1. The new dependency it creates, [claim 32: unconfirmed], is registered as A7 and reviewed in IR-3 rather than assumed.

IR-3 claim 32 is unconfirmed and P1 now rests on it — 2026-09-13 commission

The statement: [claim 32: unconfirmed] — that a material share of the addressable stress-urinary-incontinence population is commercially insured rather than Medicare.
What the check found: nothing yet. It was surfaced as an assumption and has never been attempted against a source.
Who depended on it: register row A7 — Commercial and Market Access. Before P1 nobody did; the pivot is what made it load-bearing, which is exactly the moment to register it.
What it costs: if it is false, P1 routes around Medicare and in doing so routes around the market, and the pivot is worse than the shape it replaced.
Outcome: commission
Because: unconfirmed here means unasked, not unanswerable — prevalence by payer mix for stress urinary incontinence in women 45+ is a researchable number. REQ-0003 is the commission. Recording this as an impact review rather than as a footnote to the pivot is deliberate: a pivot that creates a new load-bearing assumption and does not review it has moved the risk rather than reduced it.

IR-4 claim 21 is unconfirmed, and unsearched — 2026-09-13 commission

The statement: [claim 21: unconfirmed] — the FTO claim.
What the check found: no connector call was possible and none was made. The patent primitive takes a specific patent_id and cannot run a landscape search; the claim names no patent number and no assignee, so there was no object to pass it. The claim is unsearched, not merely unverified.
Who depended on it: register row A6 — Legal & IP, which has abstained rather than hold a position on it.
What it costs: nothing yet, and that is the trap. An unsearched FTO claim costs nothing right up until the point where it costs the programme, and the tag it carries is the same tag a searched-and-inconclusive claim would carry.
Outcome: commission
Because: this is a missing capability, not a missing fact — the repo has no landscape search. REQ-0002 asks Research for a CPC-class landscape, which is the shape of work that can answer it. Absorbing it would mean deciding FTO does not matter to a device whose differentiator is its mechanism.

IR-5 claim 6 is unconfirmed and A2 rests on it — 2026-09-13 commission

The statement: [claim 6: unconfirmed] — that a device with this intended use is eligible for 510(k) clearance and requires neither a De Novo nor a PMA.
What the check found: openFDA publishes no De Novo endpoint and this repo holds no cached copy of FDA's De Novo listing, so the Verifier could not attempt it and explicitly declined to substitute a web search. What is established — [claim 2: verified] carries submission_type_id 1, and several KPI clearances exist — establishes that the product code is a 510(k) code. It does not establish that this device, with this intended use, is substantially equivalent to any of them.
Who depended on it: register row A2 — Regulatory, which holds a conditional verdict resting on exactly this.
What it costs: a De Novo route adds 12-18 months and a new evidence burden ahead of any revenue, which moves Factor 7 as much as Factor 1. It does not change P1 — the payer-route pivot is orthogonal to the clearance route — which is why this review is separate from IR-2 rather than folded into it.
Outcome: commission
Because: this is a missing cache, not a finding — the same shape as IR-4 and the opposite of IR-2. The De Novo listing is a file FDA publishes and this repo does not yet fetch, so the honest work is to fetch it; Regulatory's [C1] names the alternative (a pre-submission meeting response) which settles the substantial-equivalence half that no listing can answer. No request is raised against Research for the first half, because engine/scripts/fetch_data_file.py is the tool for a missing cache and that is a job, not a question.

IR-6 claim 14 is unconfirmed and the recurring-revenue leg rests on it — 2026-09-13 commission

The statement: [claim 14: unconfirmed] — that pelvic floor muscle status qualifies as "musculoskeletal system status" for the purposes of billing CPT 98977, which is what puts this device on the RTM path at all.
What the check found: nothing determinative. [claim 13: unconfirmed] establishes that 98977 exists and describes device supply for remote therapeutic monitoring of the musculoskeletal system; whether pelvic floor sits inside that category is a coding question nobody has retrieved an answer to.
Who depended on it: register row A8 — Market Access, via its [C2]. It arrived in the register through that binding rather than through the front end, which is the mechanism working as designed.
What it costs: the money model has two legs, a device sale and a recurring RTM stream. If this is false the second leg closes and the opportunity becomes a one-time sale of a $350-$700 device — which is a different business, not a smaller one. Note that this cost is independent of P1: it bites the same way under Medicare and under commercial payers.
Outcome: commission
Because: unasked rather than unanswerable, and cheap to ask — CPT Assistant guidance or any payer policy naming pelvic floor within the RTM musculoskeletal category settles it. Market Access's [C2] already names that source. Folded into REQ-0001's scope rather than raised separately: it goes to the same department, against the same coding database, in the same sitting.

How the shape changed

P1 — commercially-insured first, Medicare later — 2026-09-13

Caused by: IR-2, and pointed at independently by Market Access's [C3] from the coverage side.
What changed: the population the opportunity sizes and sells to first. Medicare fee-for-service → commercially-insured women 45+, with Medicare as a later route rather than the launch route. The intended_use: frontmatter is unchanged — the eligibility language it mirrors is still the clinical gate, and Regulatory has said there is no class benefit in narrowing further.
Supersedes claims: none. This matters and is easy to get wrong: claims 16 and 17 are correct findings about Medicare fee-for-service volume and stay live as exactly that — a floor. What the pivot supersedes is the register rows A4 and A5, the dependency of this opportunity on those numbers being its market. A pivot supersedes claims only when a claim is about the shape that changed; here the claims were never wrong, the thing resting on them was.
Chairs that must re-state: Commercial, whose market model is built on the superseded base and which has no position on file yet. Market Access's [C3] already anticipates this shape, and Regulatory and Legal & IP have both said in writing that a payer-route pivot does not change their answers — which is why the pivot: bump to P1 does not show them as stale.
What we learned: *a code with a published fee schedule amount is not a covered code, and Medicare fee-for-service utilisation is a floor on volume rather than a market size.* Both are transferable to every future candidate in this domain, and both were stated confidently in the original framing. The next opportunity that sizes off a CMS utilisation figure should start from A4/A5 rather than rediscover them.

Open questions between departments

Any department may ask any other what it cannot answer itself. The answer must name an artifact that exists — a research report, a position, a dossier — never a summary.

Request	From → to	Question	State
`REQ-0001`	regulatory → market-access	Does the NCD 230.8 eligibility language reach this indication as written?	accepted
`REQ-0002`	legal-ip → research	A patent landscape over closed-loop EMG-gated pelvic floor stimulation, by CPC class (A61N1/36007, A61N1/05, A61B5/389) rather than by patent number, with live/expired status per family	open
`REQ-0003`	front-end → commercial	Payer mix for diagnosed stress urinary incontinence in US women aged 45+: what share is commercially insured rather than Medicare-primary?	open

What this is waiting on

Rebuilt from the files, not remembered: python3 engine/scripts/check_loop.py --candidate idea-001 --open

Clinical & Scientific has not stated a position
Technology & R&D has not stated a position
Operations & NPI has not stated a position
Commercial has not stated a position
REQ-0001 accepted: regulatory → market-access
REQ-0002 open: legal-ip → research
REQ-0003 open: front-end → commercial

Elsewhere

The packet All claims How the loop works

Nothing on this page is a recommendation, and no verdict here is a decision. This is a compilation of what the loop record, the position files and the request ledger say. Shortlist review and every decision past it are human-only.

Generated from knowledge-base/loop/idea-001.md, knowledge-base/positions/idea-001-*.md by engine/scripts/build_review.py on 2026-09-14. This page is a view — edit the source file and rebuild, never edit here.