idea-018 · viability 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 investor, and the operator who has to run this as a business.
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 copied from it, never adjusted here. A no-claim marker means nothing in the file speaks to the condition at all. See docs/deck-spec.md.
No number appears on this deck that is not in a claim, and there is no ask slide: a raise amount would be a fabricated financial figure.
Would have to be true: The money would have to come out of the Medicare Part B durable medical equipment benefit, billed by an enrolled DMEPOS supplier under HCPCS E0747 for a beneficiary whose chart carries the radiographic nonunion documentation — with the beneficiary's cost share as the only other source of cash, and commercial and Medicare Advantage plans following on comparable terms for the working-age share of nonunion patients.
Where it stands: The coverage instrument is the strongest thing in the file and it is verified. NCD 150.2 covers the non-invasive stimulator for "Nonunion of long bone fractures" among a closed list of indications — the policy says "covered only for the following indications" [claim 8: verified] — and defines the nonunion this candidate's intended use restates [claim 9: verified], excluding exactly the categories the intended use excludes [claim 10: verified]. The code that carries the money is not verified: E0747's existence and descriptor could not be checked in either direction because the HCPCS quarterly release is not cached [claim 13: unconfirmed]. Whether the device is paid separately rather than absorbed into another payment is likewise unchecked [claim 23: unconfirmed]. Non-Medicare payers are nowhere in the file; this deck put that on record [claim 34: unverified].
What would settle it: The CMS HCPCS Level II quarterly release for the E0747 descriptor (data/hcpcs_level_ii.csv, per data/README.md), and the DMEPOS payment rules for the code. For the payer-mix leg, published commercial medical policies for osteogenic stimulators from two or three national plans — a human retrieval, not a connector call.
If it's false: If E0747 is not the code, or the device is bundled rather than separately payable, Factor 2 (Reimbursement, the file's strongest factor at 4/5 and the only thing keeping Regulatory-plus-Reimbursement off the joint kill floor) loses its payment leg and reverts to coverage eligibility with no demonstrated route to cash.
Would have to be true: All three would have to land independently — the code would have to exist with the descriptor the file assumes, the allowance would have to be high enough to clear cost of goods and channel margin, and the DME MAC's documentation conditions would have to be ones this device's paperwork can satisfy on every claim.
Where it stands:
coverage connector has no data file here and is
code-keyed so it cannot resolve an article ID at all [claim 12: unconfirmed].
What would settle it: HCPCS quarterly release for the descriptor; DMEPOS fee schedule PUF for the allowance; and a human reading LCA A52513 directly in the CMS Medicare Coverage Database — the file names that article as the highest-value single document for anyone assessing this candidate, and no connector here can reach it.
If it's false: Coverage without payment is not a business. NCD 150.2 establishes eligibility and, as the verifier recorded, says nothing about whether payment is separate, bundled or adequate — so a low or bundled allowance takes Factor 2 down without touching a single verified claim.
Would have to be true: Four separate quantities would each have to hold: the annual US population reaching radiographically confirmed established long-bone nonunion; the share of them who are dispensed a stimulator; the allowance paid per device; and the share of that volume a new entrant can take from the incumbents.
Where it stands:
market returned unverifiable, so it stays unconfirmed permanently unless a
human attaches a licensed report [claim 17: unconfirmed].
What would settle it: One query against the CMS "Medicare DME, Devices & Supplies — by Geography and Service" release, pulling E0747 and E0748 together — it settles the volume leg and claim 18's relative-size half at once — plus the DMEPOS fee schedule PUF. Both free. The file names this as the cheapest high-value check it has, and it is still unrun because the release is not cached (docs/backlog.md 1.13).
If it's false: TAM is a hard gate: below 2/5 is an automatic kill under the rubric, regardless of how good the regulatory story is. Every one of the four legs above is currently unconfirmed, so Factor 3's draft 2/5 is what the guesses would imply, not a finding — the utilisation file can move it in either direction.
Would have to be true: Before anyone can be billed, the sponsor would have to build the device, satisfy every special control at 21 CFR 890.5870(b), obtain a 510(k) clearance against some predicate, enrol or contract DMEPOS distribution, and fund all of it with no coverage-at-authorization lane to shorten the wait.
Where it stands: The regulatory burden is now partly known and it is heavier than "commodity hardware onto an existing pathway" suggests. The special controls were read in full and (b)(1) requires clinical data demonstrating the device performs as intended, with imaging data demonstrating fusion at the treatment site; (b)(2) non-clinical performance testing; (b)(5) software verification, validation and hazard analysis; (b)(6) labeling comprehension testing [claim 3: verified]. So the 510(k) is not a bench-only submission. Worse for the calendar: no 510(k) has ever been cleared under product code LOF or LPQ — an empty openFDA query, not a query not run — so four months after the route opened there is no device cleared under 890.5870 to serve as a predicate [claim 20: verified]. The RAPID coverage-at-authorization lane is outside reach on the Generator's reading, and the file records that RAPID is itself a notice open for comment, not an operating pathway [claim 22: unconfirmed]. No claim in this file carries a cost, a burn rate, a headcount or a calendar, so no figure appears here. The three conditions that set what this costs are stated on the feasibility deck and referenced from here: predicate availability [claim 38: unverified], the clinical data set that would satisfy (b)(1) [claim 40: unverified], and the coil engineering [claim 39: unverified]. Unit cost against the allowance is [claim 41: unverified].
What would settle it: An FDA pre-submission meeting — which the file already nominates for claim 7 — asking two questions at once: is a pre-reclassification PMA device available as a predicate, and what clinical data set satisfies (b)(1) for a nonunion rather than a fusion indication. Nothing cheaper answers either.
If it's false: If there is no usable predicate, the "cheap pathway" the whole candidate rests on is a 510(k) with no predicate — a De Novo-scale evidence programme — and Factor 7 (capital intensity, drafted at 3/5 explicitly pending this) falls with Factor 1. The narrowing was bought precisely to avoid that outcome.
Would have to be true: Something other than the pathway would have to be hard to copy: a patent on the garment-integrated coil or the tamper-evident log, a documentation requirement a competitor's device cannot meet, or a supplier relationship that locks the channel.
Where it stands: The pathway is explicitly not it — the 510(k) route opened on the same day to every incumbent and every other entrant, and the file records this as the strongest argument against the Moat factor [claim 21: unconfirmed] (the factual premise is verified at [claim 1: verified] and [claim 6: verified]; what it implies for exclusivity is reasoning). The patent leg is a void, not a negative: no patent search of any kind was performed, for the waveform, the coil geometry, or the adherence-logging function [claim 26: unconfirmed], and no patent, family, assignee record or expiry date was located for any incumbent [claim 27: unconfirmed]. Unexamined is not clear, and the Stage 5 FTO kill cannot fire on it. That protection is obtainable at all is this deck's own assumption [claim 36: unverified]. The documentation leg is the entire commercial case for the one differentiating feature and the file labels it as an assumption in those words: DMEPOS suppliers and payers would have to treat a device-generated adherence log as materially useful for continued-coverage support, audit defence or denial reduction [claim 30: unconfirmed].
What would settle it: LCA A52513's documentation conditions, read by a human — the cheapest partial test of whether an adherence record carries weight with a DME MAC. Then a PatentsView or EPO OPS assignee search against Orthofix, Zimmer Biomet, Bioventus and Enovis plus a CPC search on compliance monitoring; this repo has patent but no assignee-search connector, so it needs a patent number or a human.
If it's false: Factor 5 (Moat) is already drafted at 1/5. If the adherence log turns out to be a feature nobody pays for, there is no differentiation left at all — the hardware is deliberately unremarkable, by the candidate's own rationale note, and the moat would rest on DMEPOS channel execution alone.
Would have to be true: The four named incumbents would have to leave the low-profile and adherence-logging ground unoccupied long enough for a new entrant to take it — and "nobody has cleared a device under the new regulation" would have to mean opportunity rather than an empty market.
Where it stands: The incumbents — Orthofix (Physio-Stim, Spinal-Stim), Zimmer Biomet (OrthoPak), Bioventus (EXOGEN, ultrasonic), plus Enovis at claim 27 — are named on the Generator's background knowledge, with no PMA number, K-number or GUDID identifier asserted for any of them [claim 19: unconfirmed]. The 510(k) searches that returned refuted are answering the wrong question: a PMA-marketed device does not appear in the 510(k) endpoint, and this repo has no PMA, GUDID or registration-and-listing connector. What is verified is the emptiness of the new route: zero clearances under LOF or LPQ ever, cross-checked by device name [claim 20: verified] — and the file is explicit that this is not white space in the market, because the incumbents are there under PMAs. That no incumbent ships a low-profile or adherence-logging version first is this deck's assumption [claim 37: unverified], and prescriber preference already runs their way [claim 28: unconfirmed].
What would settle it: A PMA/P-number or GUDID lookup, or FDA registration-and-listing data, to establish who is actually marketing what — none of which exists in this repo. Then openFDA recall and enforcement records for the named firms, and a public-filings or press sweep on recent launches, both free and quick.
If it's false: If an incumbent ships the same two differentiators first, Factor 8 (Competitive intensity, drafted at 1/5) does not move but Factor 5 goes to zero in substance: the candidate becomes a fourth brand of commodity PEMF with no channel and no brand.
Would have to be true: One free CMS query would have to be capable of moving the money question from assumption to fact — the E0747 and E0748 allowed-services counts and the DMEPOS purchase allowance — and it would have to come back at a size that justifies the narrowing.
Where it stands: The query is named in the file three times over and has never been run, for a tooling reason rather than a judgement one: the procedures connector's cached release carries essentially no E-prefixed codes, and the file records "Nothing was looked up in either direction" [claim 15: unconfirmed], [claim 18: unconfirmed]. The allowance needs data/hcpcs_payment_rates.csv, which does not exist here [claim 16: unconfirmed]. Everything else on this deck is more expensive: the pre-submission (Slide 4), the interviews (Slide 5), the licensed epidemiology (Slide 3).
What would settle it: Cache the CMS "Medicare DME, Devices & Supplies — by Geography and Service" release and the DMEPOS fee schedule PUF, then run E0747 and E0748 in one pass. This is a connector and data-file work item (docs/backlog.md 1.13), not a research programme.
If it's false: If even the utilisation file cannot resolve E0747 volume, the TAM has no checkable leg at all — claim 17's epidemiology is permanently unconfirmed by design — and Factor 3 stays unverifiable rather than low, which routes the candidate to queued-research rather than to any conclusion about its size.
Every condition above with nothing verified behind it — the [no claim] markers, plus the unverified claims this deck itself put on record. Read this slide first.
apc has no
data file here either.
The count that matters: the money slides rest on fourteen lines, none of them verified. What is verified on this deck is coverage eligibility (claims 8-11) and two facts that point downward — the special controls require clinical data, and no predicate exists yet under the new regulation (claims 3, 20).
If only one thing from this chair could be checked: Whether the Medicare line this candidate is aimed at is big enough to be worth entering — annual E0747 allowed services at the DMEPOS purchase allowance [claim 15: unconfirmed], [claim 16: unconfirmed].
This one rather than the others because it is the only leg of the money case that is free, fast, and decisive in both directions, and because the candidate's central move — narrowing the indication to fit NCD 150.2 — was made in exchange for exactly this line. The regulatory questions (predicate, special controls) change what it costs to get there; this one decides whether there is a "there". The file itself nominates it as its own first check and it is still unrun because a data file is missing, not because anyone judged it hard.
Naming it is not a recommendation, a gate, or a kill.