idea-017 · 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 would have to run this as a Medicare DME business — selling through suppliers who bill a capped monthly rental, against two entrenched incumbents, on a coverage instrument CMS has already written.
How to read this: every slide is a condition, not a conclusion. A bracketed claim reference points into knowledge-base/candidates/idea-017.md with the tag it carries there; a no-claim marker means nothing in the file speaks to this. No number appears here that isn't in a claim, and there is no ask slide. See docs/deck-spec.md.
The shape of this file, stated once. Its reimbursement premise is unusually strong in kind and unusually empty in quantity: CMS has already issued a national coverage determination for exactly this population, and the file retrieved not one of the dollar amounts the business runs on. Three of this deck's conditions had nothing behind them before this build and are now write-backs (claims 34-36).
Would have to be true: Medicare Part B's durable medical equipment benefit would have to pay for this, billed by a DME supplier as a monthly capped rental under HCPCS E0471 — meaning the customer who has to be sold is the DME supplier, not the patient and not the prescribing practice.
Where it stands: The payer and the instrument are established; the code linkage and the payment category are not. CMS established NCD 240.9, "Noninvasive Positive Pressure Ventilation (NIPPV) in the Home for the Treatment of Chronic Respiratory Failure (CRF) Consequent to Chronic Obstructive Pulmonary Disease (COPD)", effective 2025-06-09 [claim 1: verified], carried into the manual by transmittal R13808 with implementation 2025-10-22 [claim 3: verified]. The code-to-NCD linkage is weaker than the file's framing suggests: the NCD does set national criteria for respiratory assist devices and home mechanical ventilators, but its text names no HCPCS code at all — neither E0470 nor E0471 appears in it, so that linkage is this repo's own and not CMS's wording; and section D expressly preserves MAC discretion rather than replacing it [claim 4: unconfirmed]. The capped-rental payment category could not be checked in either direction: the 2006 Federal Register rule that classified these devices uses a legacy document number the policy connector rejects before querying [claim 8: unconfirmed]. That the DME channel would carry it at all is an unresearched assumption [claim 28: unconfirmed].
What would settle it: A human retrieval of the 2006 rule "Medicare Program; Payment for Respiratory Assist Devices With Bi-Level Capability and a Backup Rate" (or a connector fix for pre-2007 FR document numbers), plus the DMEPOS fee schedule file, which carries the payment category alongside the allowable.
If it's false: If E0471 is not the code, or not capped rental, the revenue model is not a monthly rental at all and the entire "first revenue does not wait on a new code" argument that holds up the capital-intensity factor goes with it.
Would have to be true: The code would have to exist with a descriptor this device fits; the allowable would have to be a number large enough to build a business on; and the coverage policy would have to permit the specific thing this device does — settings set and documented from a home overnight tcCO2 recording.
Where it stands:
What would settle it: Three free retrievals a human can do in an afternoon — the CMS HCPCS Level II release for the E0471 descriptor, the DMEPOS fee schedule for the allowable and payment category, and LCD L33800 with article A52517 (and the proposed DL33800 and response-to-comments A58822) for the documentation requirements. None is reachable by any connector here.
If it's false: If the documentation half fails, the device's novel function is unbillable and the reimbursement factor's premise — that CMS has already performed the site-of-care shift — becomes a statement about a therapy category rather than about this product. If the allowable is too small, see Slide 7.
Would have to be true: Population times hypercapnic-CRF rate times reachable share times annual revenue per enrolled patient would have to produce a number worth building for — and each of the four would have to be a retrieved figure rather than an estimate.
Where it stands: The decomposition is sound and not one of the four legs carries a retrieved number. The file asserts no market size of any kind, deliberately.
What would settle it: Two free CMS downloads close half of it — "Medicare Durable Medical Equipment, Devices & Supplies by Geography and Service" for E0471 allowed services, and the DMEPOS fee schedule for the allowable. The prevalence legs have no connector reach and need a human-attached primary table.
If it's false: TAM carries a hard kill floor at 2/5 in the rubric, and it is currently scored on what the four claims would imply if all four held rather than on any retrieved figure. A small hypercapnic-CRF share against a capped monthly rental is the arithmetic that would find the floor.
Would have to be true: A two-subsystem hardware product — a bi-level ventilator with back-up rate, plus a reusable heated sensor with a consumable element, plus the software that computes a settings proposal — would have to reach a 510(k) clearance and a first billed rental without waiting on a new code or a new coverage decision.
Where it stands: The regulatory half of the sequence has real anchors and the cost-and- calendar half has nothing. Both subsystem categories are established Class II device types with identified product codes: cutaneous CO2 monitors are Class II under special controls at 21 CFR 868.2480, product code LKD [claim 17: verified], and continuous ventilators sit at 21 CFR 868.5895, product code MNS, with the Philips BiPAP AVAPS cleared under it as K102465 [claim 18: verified]. What has no predicate is the combination: no cleared device combining the two functions was identified, and the file is explicit that this is an absence-of-evidence result rather than white space [claim 19: unconfirmed]. The claim that FDA would accept a 510(k) for the combined device without a De Novo or a clinical study is this deck's write-back and is stated at length on the feasibility deck's Slide 3 [claim 38: unverified]. No claim in this file carries a development cost, a headcount, a burn rate or a calendar to first revenue [no claim] — and none is estimated here. On the coverage-timing side, the one policy thread that might have accelerated things does not apply: the RAPID pathway is limited to Class II and III FDA Breakthrough-designated devices [claim 21: verified], and its eligibility list opens with "Not already the subject of a controlling Medicare NCD" immediately before the separately-payable criterion the Generator cited [claim 20: verified] — so on the document's own words this candidate appears excluded by the very NCD its premise rests on, and no timing tailwind should be credited here.
What would settle it: A pre-submission meeting with FDA on the combined-function question — a human act, not a database lookup — is the only thing that settles the pathway. The cost and calendar need an engineering-led build estimate, which is a department position rather than a source retrieval.
If it's false: If the combined device needs a De Novo with a clinical study, the capital-intensity factor's whole argument — that first revenue does not wait on a new code or a new coverage decision — survives on the reimbursement side and dies on the regulatory side, and the calendar lengthens by the duration of a trial nobody has budgeted.
Would have to be true: Something specific would have to stop ResMed or Philips Respironics from adding a transcutaneous CO2 recording channel to a bi-level platform they already have cleared — a patent, an exclusive sensor supply, a data asset, or a coding requirement they cannot meet.
Where it stands: Nothing in the file identifies any such mechanism, and the IP surface is unexamined rather than clear. No patent search was performed and no patent number, family or assignee list is named anywhere in this candidate; the file states the field is unexamined, not clear, and this pass confirmed it is still unexamined — structurally so, because there is no patent search connector in this repo, only a lookup that requires a specific patent number [claim 25: unconfirmed]. The expectation that the ~1990 foundational bi-level-with-back-up-rate patents have expired is inference from the modality's age, and the file notes that even a favourable lookup would date one document rather than clear the modality [claim 26: unconfirmed]. The proposed moat is the integration itself, and whether that integration is patentable over the incumbents' existing filings is this deck's write-back [claim 35: unverified]. Two named risk surfaces are on the record without having been searched: volume-assured/target-ventilation control algorithms of the iVAPS and AVAPS kind, and tcCO2 sensor head construction, heating and calibration [claim 25: unconfirmed].
What would settle it: A patent search on those two surfaces against the ResMed, Philips Respironics, Sentec AG, Radiometer Medical and Perimed AB assignee portfolios, via PatentsView or EPO OPS. The file already names the starting points a searcher needs: K161492 for the ResMed platform, K102465 for the Philips one, and product code LKD for the sensor surface. Cost is low once an API key is confirmed available in this environment, which it has not been.
If it's false: FTO carries the rubric's only automatic kill, and it is currently unexamined. If a live patent covers either named surface with no workaround, this is not a weak moat — it is a gate condition. Read alongside the feasibility deck's Slide 7, which states the same claims as a design constraint.
Would have to be true: The incumbent field would have to be occupied in a way that leaves an opening — and the opening would have to be structural rather than a temporary dislocation that could close before this product ships.
Where it stands: The field is crowded and the one opening is of unknown current status. Both halves of this device are shipped separately today by well-capitalised incumbents: Philips' BiPAP AVAPS is cleared as K102465 under product code MNS [claim 18: verified], and on the sensor side SenTec's Digital Monitoring System is cleared as K071672 with the V-Sign sensor named explicitly in K151329 and the current tCOM+ as K232516, while ResMed's marketed "AirCurve 10 ST-A with iVAPS" does not resolve under that name in openFDA — the nearest record is K161492, "Juno VPAP ST-A", and the brand-to-clearance mapping is left for a human [claim 24: unconfirmed]. Product-code-keyed scans show the two halves cleared under disjoint codes with no overlap: 43 MNS clearances since 2000 with no device name mentioning CO2, and 9 LKD clearances all standalone monitors — consistent with the claim that no combined device is cleared, and explicitly not evidence for it, since a combined device could be cleared under either code without a name a scan would catch [claim 19: unconfirmed]. The supposed opening is the 2024 Philips consent decree restricting US sales of new CPAP and bi-level sleep-therapy devices [claim 23: unconfirmed] — and the file is blunt that whether those restrictions are still in force in September 2026 is unknown to this file, because no connector here reaches a court docket, a DOJ filing or an FDA enforcement action; nothing was queried. What is verified about Philips is the older event: the 2021-06-14 PE-PUR foam recall notification, confirmed across six openFDA recall records spanning ventilators, bi-level and CPAP devices, all still "Open, Classified" [claim 22: verified].
What would settle it: A human reading the court docket or Philips' current investor filings for the decree's status — named in the candidate's own Scores table as the one item that changes the competitive read — plus an openFDA brand-to-clearance resolution for the ResMed AirCurve naming question.
If it's false: If the decree has been lifted, the only opening the competitive-intensity factor credits disappears and the space reads as two entrenched incumbents one product-line extension away from the proposed differentiator. A two-year-old enforcement dislocation is the kind of fact most likely to have moved, and nothing in this file has looked.
Would have to be true: The E0471 monthly capped-rental allowable would have to cover the ventilator, the reusable sensor and its consumables, and the DME supplier's setup and service burden, with margin left for the manufacturer — and patients would have to stay adherent enough for those monthly rentals to keep being paid.
Where it stands: Neither half can be stated numerically from this file. The allowable was not retrieved because the payment cache is absent [claim 10: unconfirmed], and the file separates the two halves of the supplier question precisely: the preference half is permanently unconfirmed absent primary research, and the quantitative half — does the extra service burden fit inside the allowable — is blocked on the missing file, making it a two-step item: retrieve the allowable first, then take supplier margin to primary research [claim 28: unconfirmed]. On the persistence of the revenue, the adherence bar is a payment condition and not only a clinical one: NCD 240.9 requires use "at least 4 hours per 24-hour period, on at least 70% of days in a 30-day period", re-established at 6 months and again at 7-12 months [claim 29: unconfirmed], and whether real patients clear it often enough for rentals to continue through the capped period is this deck's write-back [claim 36: unverified]. The one side of the three-sided payment comparison that now carries a number is the study being displaced: CPT 95811 pays $711.30 globally, with the technical component at $586.76 — roughly 82% of the payment is the facility-side money — while the DMEPOS allowable and the inpatient side remain blocked on absent caches, with MS-DRGs 190/191/192 still unidentified [claim 14: unconfirmed].
What would settle it: One free download — the CMS DMEPOS fee schedule for E0471. It is the single cheapest item in this file and it unblocks the price leg of TAM (Slide 3), the patient coinsurance on the desirability deck's Slide 4, and the supplier-margin question above. After it lands, supplier interviews price the service burden against it.
If it's false: If the allowable does not cover the extra sensor and its service, the DME channel will not carry the product, and there is no alternative channel in this file — Medicare RADs are dispensed and billed by suppliers, not sold to patients or practices. This is this deck's load-bearing condition; see Slide 9.
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.
policy connector refuses.
The honest measure is not the two no-claim markers. It is that of the sixteen conditions above, the only ones with verified claims behind them are the existence of the coverage instrument, the two device classifications, the RAPID exclusion and the Philips recall — the structure of the opportunity. Every number the business would run on is unconfirmed or unverified, and most of them are blocked on files that are free to download.
If only one thing from this chair could be checked: whether the Medicare DMEPOS monthly capped-rental allowable for E0471 covers the ventilator, the reusable tcCO2 sensor and its consumables, and the DME supplier's setup and service burden with margin left over — a number no claim in this file carries [claim 10: unconfirmed].
This one rather than the others because it is simultaneously the price leg of TAM, the reason a supplier would or would not stock the device, the patient's coinsurance exposure on the desirability deck, and the constraint on how expensive the sensor is allowed to be on the feasibility deck. It is also, uniquely among the big unknowns here, free: the DMEPOS fee schedule is a public CMS download, and the only thing standing between this file and the number is an absent local cache. Every other viability unknown costs either a patent search, a pre-submission meeting, or primary research.
Naming it is not a recommendation, a gate, or a kill.