idea-017 · 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: engineering. The people who would have to build a home bi-level ventilator with back-up rate, a reusable heated transcutaneous CO2 sensor, and software that turns a night's trace into a settings proposal a practitioner approves — and produce the evidence a 510(k) (premarket notification) reviewer would ask for.
How to read this: every slide is a condition, not a conclusion. A bracketed claim reference points into knowledge-base/candidates/idea-017.md: the number is the claim's position in its ## Claims list and the tag is the one it carries there, copied and never adjusted. A no-claim marker means nothing in the file speaks to the condition. See docs/deck-spec.md.
What changed since the last build. Every condition this deck wrote back on 2026-09-20 (claims 37-40) has now been through a Verifier pass and is unconfirmed — none contradicted, none confirmed. Two findings sharpen the engineering read: 21 CFR (Title 21 of the Code of Federal Regulations) 868.2480 frames a cutaneous CO2 monitor as an adjunct that tracks relative change, and the 2020 ATS guideline's titration target is arterial PaCO2. Together they move the central engineering question from "did the trial use tcCO2 (transcutaneous carbon dioxide)" to "can a home tcCO2 sensor carry a titration target at all". One condition had nothing behind it and is now claim 45unverified.
Would have to be true: The device would have to deliver bi-level pressure with a fixed back-up rate across at least the NCD (national coverage determination)'s high-intensity floor and the pressures HOT-HMV actually used; record overnight tcCO2 from an ear or chest sensor; store both traces together; compute a proposed settings change against a practitioner-set CO2 target; and change nothing without practitioner approval, as the intended_use: states.
Where it stands: The floor is IPAP ≥ 15 cm H2O with a back-up rate of at least 14 breaths per minute, from the NCD's own definition; the file's original "above 20 cm H2O" is claim 5refuted. The upper working range is trial practice: median IPAP 24 cm H2O (IQR 22-26), EPAP 4, back-up rate 14 claim 15verified. The pressure ceiling the device must reach, the sensor's operating temperature, recording duration and the form of the "CO2 target" are blanks the file does not fill — those are design inputs, stated on Slides 4-5 as conditions (claims 39, 40), not specified here.
What would settle it: A design-input document that takes the NCD floor and the trial IQR as its pressure range and states the sensor and software requirements as testable numbers — an engineering act, not a lookup.
If it's false: A device that cannot reach the trial's range cannot deliver the therapy the evidence is about; a device that changes settings on its own leaves the 510(k) route (Slide 3).
Would have to be true: Overnight tcCO2 recorded at home with a reusable ear or chest sensor would have to agree with arterial PaCO2 closely enough, and drift little enough across a night, to serve as the titration target — not only as an indicator of relative change — and titrating to it would have to be what produces the benefit HOT-HMV measured.
Where it stands: The agreement half has no evidence in either direction and the regulation frames the device narrower than the condition needs: 21 CFR 868.2480 identifies a cutaneous CO2 monitor as intended "to monitor relative changes ... as an adjunct to arterial carbon dioxide tension measurement" claim 37unconfirmed, with the Class II (FDA device class II) status itself confirmed claim 17verified. The ATS 2020 guideline, a pointer only, suggests "targeted normalization of PaCO2" — an arterial target claim 32unconfirmed. The trial-protocol half is a genuine source silence: the HOT-HMV structured abstract says nothing about tcCO2 or the titration target claim 16unconfirmed, though the trial's result itself is confirmed claim 15verified.
What would settle it: Published method-comparison studies of overnight tcCO2 against arterial PaCO2 in hypercapnic COPD, appraised by a human (none were searched); the HOT-HMV methods section for PMID (PubMed identifier) 28528348 (free to a human, paywalled to every connector here).
If it's false: The sensor reports a trend nobody may titrate to; the recommendation software has no valid target, and the Mechanism factor loses the premise the whole device is built on. This is this deck's load-bearing condition — see Slide 9.
Would have to be true: FDA (US Food and Drug Administration) would have to take the combined device on a 510(k) — the ventilator inside the MNS space, the recording inside the LKD space, the novel function confined to software that recommends rather than acts — rather than read the combination as a new device type requiring De Novo (De Novo classification request) and a clinical study; and the build would have to meet both halves' controls.
Where it stands: Each half has a confirmed home. 21 CFR 868.5895 is Class II (performance standards), product code MNS, with K102465 (an FDA 510(k) clearance number) (BiPAP AVAPS) as a cleared device under it — though K102465's indication names neither CRF consequent to COPD nor any CO2 function claim 18verified. 21 CFR 868.2480 is Class II (special controls), product code LKD, whose special control is an FDA guidance document whose text nobody here has read claim 17verified. No combined-function predicate was found, which is an absence-of-evidence result claim 19unconfirmed; the cached De Novo listing has no grant matching "ventilator" or "carbon dioxide", so there is no De Novo precedent either way claim 38unconfirmed. What the two controls require as bench tests, software documentation and human-factors work is unread — a blank inside claims 17 and 18, not a finding.
What would settle it: An FDA pre-submission on the combined device and its advisory-only software function (a human act); a human read of the PcCO2 special-controls guidance and the performance standards 868.5895 invokes.
If it's false: Route 3 in device_class: governs: De Novo with a clinical study, and every bench plan on Slide 5 becomes a subset of a larger evidence package.
Would have to be true: The reusable sensor head — a heated electrochemical element held against ear or chest skin — would have to hold calibration across an unattended overnight home recording without mid-session recalibration or membrane change, at a skin temperature that causes no thermal injury to elderly COPD patients who may not reposition themselves.
Where it stands: Nothing in the file bears on it; the claim records that no connector covers sensor calibration stability or thermal safety and nothing was queried claim 39unconfirmed. The existing LKD clearances (SenTec K071672, K151329, K232516) are the named places to look, and SenTec's cleared home indication is "if under clinical supervision" claim 24unconfirmed.
What would settle it: A human read of the LKD predicate 510(k) summaries for stated monitoring duration, calibration interval and sensor temperature; a MAUDE (Manufacturer and User Facility Device Experience) search for burn reports under product code LKD; then a bench drift test over a full night.
If it's false: An unattended overnight home recording needs a design the incumbents' cleared sensors do not already demonstrate, and the product's one novel measurement channel is the part with the least precedent.
Would have to be true: Before any patient is involved, a bench rig would have to show that the titration software never proposes a settings change from a corrupted trace — detachment, poor skin contact and artefact are detected and shown to the practitioner as a rejection rather than averaged into a number they sign — and that the sensor holds calibration over a simulated night (Slide 4).
Where it stands: Trace rejection is a design requirement for software that does not exist, recorded so it can enter a verification plan claim 40unconfirmed. Calibration hold is claim 39unconfirmed. Neither has any evidence, which is the expected state for an unbuilt device.
What would settle it: A bench rig with injected detachment, contact-loss and motion artefact against the software's rejection logic, and a drift test against a reference gas over a full night. Engineering work; a human team runs it. Anything past the bench — a patient or an animal — is a Tier 3 decision outside this deck.
If it's false: The practitioner is asked to approve a proposal that may rest on a bad trace, which is the device's primary safety hazard rather than a performance shortfall.
Would have to be true: The system would have to be a durable rental asset a DME (durable medical equipment) supplier can clean and reprocess between successive patients, with the consumable sensor element replaced by the supplier's staff in the patient's home and no manufacturer field service; and a patient or caregiver would have to place the sensor unaided on prescribed nights.
Where it stands: Reprocessing and supplier-side consumable replacement had nothing behind them before this build claim 45unverified. Supplier willingness to stock and service the sensor is an assumption on record claim 28unconfirmed, and the money half of it is on the viability deck's Slide 7. Unaided home application — the human-factors reality of a bedroom at night — is the desirability deck's Slide 6 condition claim 33unconfirmed.
What would settle it: A reprocessing validation of the combined system and a supplier-run in-home consumable change, observed; the ten-patient in-home usability study named on the desirability deck, which is also human-factors evidence a 510(k) would ask for.
If it's false: The device does not fit the channel Medicare pays through — a rental asset that needs the manufacturer in the home is a service business the E0471 (a HCPCS code for durable medical equipment) allowable was not built for.
Would have to be true: The design would have to be practisable without a licence — in particular the volume-assured/target-ventilation control approach and the sensor head's construction, heating and calibration would have to be free to practise or designable around without giving up the function.
Where it stands: Unexamined, not clear, and structurally so: no patent search was performed on either named surface, and no search-capable patent connector exists in this repo claim 25unconfirmed. That the ~1990 bi-level-with-back-up-rate foundational patents have expired is inference from the modality's age, and even a dated patent record would not be an FTO (freedom to operate) conclusion claim 26unconfirmed. The integration claim's CPC sweep returned 12 families, none titled as this integration; the closest title is EP3238768A1 "System for delivering oxygen and preventing hypercapnia" (2017), not yet read claim 35unconfirmed.
What would settle it: Assignee sweeps of ResMed, Philips Respironics and SenTec on both surfaces, patent <id> on EP3238768A1 and any close family, and a draft FTO read by patent counsel — FTO is a legal conclusion no connector returns.
If it's false: FTO carries the rubric's only automatic kill; a blocking patent on the sensor head would also leave the Slide 4 unknown to be re-engineered around.
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.
One reference on this deck is refuted, and it relaxes the spec: the NCD's high-intensity floor is IPAP ≥ 15 cm H2O, not above 20 claim 5refuted.
Eleven of this deck's conditions rest on unverified or unconfirmed claims. The verified anchors (claims 15, 17, 18) establish the two regulatory homes and the trial's pressure range — the envelope the device has to fit — and none of them bears on whether the sensor can carry a titration target.
If only one thing from this chair could be checked: whether overnight transcutaneous CO2 recorded at home with a reusable ear or chest sensor agrees with arterial PaCO2 closely enough, and drifts little enough across the night, to serve as the titration target for high-intensity noninvasive ventilation rather than only as an adjunct indicator of relative change claim 37unconfirmed.
This one rather than the others because it is the physical premise every other engineering slide assumes: the software computes against this signal, the sensor's hardest property is whether it holds it, and the 510(k) framing depends on what the signal is claimed to mean. It is also where the file has moved since the last build — the regulation describes the device type as an adjunct tracking relative change, and the guideline's target is arterial — while the trial-protocol question (claim 16unconfirmed) would, even answered yes, only show what a supervised trial did, not what an unattended home sensor can do.
Naming it is not a recommendation, a gate, or a kill.
_shared/plain-language.md