idea-005 · 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 and sleep scientists who have to make an EEG-free montage score a child's night the way a lab would.
How to read this: every slide is a condition, not a conclusion. A bracketed claim reference points into knowledge-base/candidates/idea-005.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 — this deck's central statement is now wrong and has been replaced. That build said this candidate "has no verified claims at all". It now has seven [claim 26: verified] through [claim 32: verified]. Two of them bear directly on this deck's load-bearing condition, and the distinction between them — whether the montage included EEG — is the most important reading on this page. Rebuilt from the file, not patched.
Would have to be true: An unattended recorder would have to capture nasal pressure and thermistor airflow, thoracoabdominal effort, pulse oximetry and body position on a child aged 5-17, and score it against pediatric AASM criteria — obstructive events over roughly two breaths, pediatric AHI thresholds — rather than adult ones.
Where it stands: No channel specification or agreement target is stated as a claim [no claim], so this deck states none. The regulatory container is verified: openFDA's classification record ties product code MNR — Ventilatory Effort Recorder, the launch product code — to its class and regulation [claim 26: verified], from a roster of 38 clearances assembled while refuting the file's premise [claim 3: refuted]. Those cleared devices' own specifications are the channel set already accepted in this class.
What would settle it: The AASM pediatric scoring manual for the criteria; the cleared MNR devices' own specifications and indications-for-use text [claim 3: refuted], [claim 26: verified].
If it's false: There is no acceptance criterion for the validation study that is the candidate's entire critical path [claim 12: unconfirmed] — you cannot power a study against an undefined endpoint.
Would have to be true: A montage without EEG would have to detect pediatric obstructive events well enough to be trusted — despite pediatric OSA presenting at lower AHI, with more partial obstruction and more arousal-based events than the adult disease this hardware was designed around.
Where it stands: This is the slide the new verified evidence transformed, and the reading requires care because two verified claims point in opposite directions. In favour: a 2026 systematic review of pediatric HSAT-versus-PSG concordance reports that in one of the largest simultaneous studies, mean AHI was 2.5±3 by HSAT versus 2.7±2.9 by PSG, p>0.05 [claim 30: verified] — close agreement on an EEG-free modality, which is exactly this candidate's architecture. Against: the same review found only 3 of 9 included studies used simultaneous testing [claim 30: verified], so the evidence base is three studies deep. And the strongest home-testing result in the file is not an EEG-free result at all: the Australian cohort's 6.6% false-positive and 3% false-negative rates were obtained with home Type-2 polysomnography [claim 31: verified], a montage that includes EEG. That distinction is the whole feasibility question. It is also precisely the AASM's stated objection — its 2017 position paper does not recommend HSAT for pediatric OSA diagnosis, citing the inability to identify literature on devices that monitor CO2 or detect arousals [claim 28: verified], and arousal detection is what EEG provides. The candidate's own dominant risk is unchanged and now well-framed: false negatives [claim 11: unconfirmed], on a montage that omits the channel the field says is critical [claim 22: unconfirmed].
What would settle it: Read the three simultaneous studies inside [claim 30: verified] for their montages, thresholds and agreement statistics — this determines whether the candidate's prospective study is a replication or a first, and it costs an afternoon. Then the candidate's own concurrent-PSG study, which is the only thing that settles it for this device.
If it's false: The device systematically misses the children it exists to find, and no amount of sensor quality fixes a denominator problem. Note that the AAP technical report already found most diagnostic screening tests in this condition had low sensitivity and specificity [claim 29: verified] — the bar this device must clear is one many have not.
Would have to be true: A 510(k) against an adult HSAT predicate would have to be acceptable — and the candidate itself says that is at best even odds.
Where it stands: Substantially better than the last build could report, and the good news came from a refutation. The premise that pediatric HSAT clearances are scarce is refuted — a cleared device states verbatim that "both pediatric and adult patients may be tested", within a roster of 38 MNR clearances over 2020-2026 [claim 3: refuted] — and the classification record is verified [claim 26: verified]. So the pediatric-permitting pathway is demonstrably walkable; several parties have walked it. What that costs the candidate is its white-space argument, which is the viability deck's problem.
What would settle it: An FDA pre-submission on the pediatric indication and its scoring criteria; the full IFU text of the pediatric-permitting clearances, read for how they handled pediatric scoring — the roster already exists in the file.
If it's false: A De Novo lands in the critical path and the capital claim [claim 12: unconfirmed] is wrong by roughly the order of magnitude the frontmatter names. The refutation makes this less likely than the last build assumed.
Would have to be true: The pediatric scoring algorithm would have to agree with concurrent in-lab PSG at pediatric thresholds — where a handful of events separates normal from treat-this-child.
Where it stands: Unbacked for this device [claim 22: unconfirmed], with the field's evidence base now measured and thin: three simultaneous studies across a whole systematic review [claim 30: verified]. The one large simultaneous comparison inside it agrees closely [claim 30: verified] — at mean AHI values of 2.5 and 2.7, which is near the pediatric treatment threshold where the decision actually gets made, so the agreement is being tested in the right range. Whether that generalises is what three studies cannot establish.
What would settle it: Retrospective scoring of existing paired pediatric PSG recordings — strip the EEG channels from archived studies and score the remainder as if it were an unattended montage. That is a dataset question, not a device question, and it can be answered long before a prospective study is funded. It also directly addresses the AASM's arousal objection [claim 28: verified], by measuring what is lost when the arousal channel is removed from the same night.
If it's false: Everything downstream — the study, the clearance, the coverage campaign — is spent on a device that cannot support the decision it is bought for.
Would have to be true: Fit, signal quality and dropout would have to be characterised on children before a several-hundred-child multi-site study is commissioned.
Where it stands: Sizing across the age range and oximetry signal quality under motion are both on record and unchecked [claim 23: unconfirmed], [claim 24: unconfirmed]. The published evidence suggests these are the smaller risks: every home study attempted in the 47-child Australian cohort produced an acceptable recording, on a heavier montage than this one [claim 31: verified]. The ordering that follows is clear — the retrospective rescoring on Slide 4 is the first thing to do, because it addresses the risk the evidence says is real.
What would settle it: A small in-lab feasibility cohort wearing both montages, measuring fit, dropout and signal quality — before enrolment at scale. And the retrospective rescoring, which needs no subjects at all.
If it's false: The pivotal study fails on technical dropout rather than on agreement, which is an expensive and uninformative way to fail.
Would have to be true: One device configuration would have to fit a range of bodies from five-year-old to adolescent, be applied by a parent, and upload a scorable record.
Where it stands: Sizing is unchecked [claim 23: unconfirmed] and parent application is unchecked for this device [claim 17: unconfirmed], though the Australian cohort spanned ages 5-16 with acceptable recordings throughout [claim 31: verified]. One quality signal is on record and verified: a recall was issued against a marketed MNR recorder — Z-0031-2022, Nox Medical Nox T3s, status Terminated, initiated 2021-09-07 [claim 27: verified]. That is a single terminated recall in this product class, not a pattern, and it is the kind of adjacent fact worth knowing before assuming the hardware category is trouble-free. Data upload, logistics and reprocessing remain absent from the file [no claim].
What would settle it: Anthropometric data across the age range against the belt and probe designs; the cleared MNR devices' own approach to sizing and reprocessing.
If it's false: The cost per study rises above the price assumption [claim 6: unconfirmed], and the economic argument against the in-lab night narrows.
Would have to be true: A pediatric-indicated multi-channel home recorder and its scoring algorithm would have to be buildable without infringing live art — and someone would have to have looked.
Where it stands: Nobody has looked [claim 25: unconfirmed] — unexamined, not clear. The refutation makes the prior worse rather than better: 38 MNR clearances over six years [claim 3: refuted] describes a populated field, and populated fields have art.
What would settle it: A patent landscape on pediatric sleep-event scoring and on home respiratory-effort recorders, starting from the assignees behind the MNR roster the file already holds; then counsel.
If it's false: Automatic kill under the Stage 5 gate — and it would land on the one asset the company would have spent mid-single-digit millions to create [claim 12: unconfirmed].
Every condition above with nothing verified behind it.
One asymmetry is worth carrying to Stage 6 and is stronger now than at the last build: the refutation that damaged this candidate's premise [claim 3: refuted] simultaneously supplied its regulatory identification — product code, class and regulation, from 38 clearances [claim 26: verified]. And the seven verified claims that followed are, unusually for this repo, mostly clinical rather than regulatory. This is the best-evidenced candidate in the portfolio on the questions that decide whether the device works. That the evidence is mixed is a finding, not a gap.
If only one thing from this chair could be checked: whether an EEG-free montage can agree with concurrent PSG at pediatric AHI thresholds well enough for a treatment decision [claim 22: unconfirmed].
It is the product, and it survives the week's evidence as the load-bearing condition — but it is no longer an open question in an empty field. Three things now bracket it. There is favourable simultaneous evidence at the thresholds that matter [claim 30: verified]. There are only three such studies [claim 30: verified]. And the best home-testing result in the file was obtained with EEG [claim 31: verified], on the same channel whose absence the AASM cites as its reason for not recommending HSAT in children [claim 28: verified].
So the question has sharpened from "does this work" to "does removing the arousal channel cost enough agreement to matter, at AHI values near 2.5". That is answerable, and the rare case where the expensive answer has a cheap precursor: archived pediatric PSG recordings can be rescored with the EEG channels removed, which approximates the device's view of the same night for the cost of analysis time rather than a multi-site study [claim 12: unconfirmed].
Naming it is not a recommendation, a gate, or a kill.