idea-004 · 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 who have to hold a specified irradiance at the skin of a moving newborn, in a garment, on battery, and measure what they delivered.
How to read this: every slide is a condition, not a conclusion. A bracketed claim reference points into knowledge-base/candidates/idea-004.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 2026-08-25 build: a first-in-human pilot of a wearable phototherapy device now sits in the file [claim 41: unconfirmed], the LBI clearance pool is enumerated with two wearable entrants named [claim 47: unconfirmed], an incumbent recall in a wear-and-flex failure mode is on record [claim 46: unconfirmed], and every deck-surfaced bench condition went through the Verifier and came back unconfirmed. Rebuilt from the file, not patched.
Would have to be true: A wrap would have to hold a specified spectral irradiance, in the ~450-475 nm band, over a specified fraction of body surface area, for the length of a prescribed episode, on battery, on a moving newborn — and hold it across the size range the population actually spans.
Where it stands: No irradiance or coverage specification is stated as a claim [claim 26: unconfirmed], so this deck states none. The size-range condition is separately on record and unchecked [claim 35: unconfirmed]. What the file gained is a set of devices that have already cleared this bar: two wrap/wearable garment entrants cleared in 2025-2026, BiliWrap (K243372) and SnugLit (K251308), within a pool of 10 LBI clearances over ten years [claim 47: unconfirmed]. Their summaries are the nearest thing to a published specification this candidate can read.
What would settle it: The 2022 AAP guideline's irradiance-at-the-skin and body-surface-area specification, plus IEC 60601-2-50, read by a human — both published documents outside this repo's connector set. Then phantom testing at both ends of the size range.
If it's false: There is no design spec, and every downstream slide is an opinion — this is the one slide whose blank cannot be worked around.
Would have to be true: Phototherapy works; the question is whether this delivery format delivers enough of it. Irradiance falls with distance and with reduced coverage, so a garment would have to hold both, at home, for hours, unsupervised.
Where it stands: The first direct evidence on the format arrived since the last build and it is a warning. In a first-in-human pilot of a wearable phototherapy device in 12 near-term/term neonates, phototherapy completed within 48 hours in 83% (10 of 12) — and 2 of 12 were switched to conventional intensive phototherapy because bilirubin kept rising despite wearable treatment [claim 41: unconfirmed]. That is a 12-infant study of a different device, unconfirmed because nothing here can appraise it, and it is nonetheless the exact failure mode the file names as this candidate's device-specific risk [claim 14: unconfirmed]: under-dosing. The outcome literature around it is thin by its own account — a pooled (re)admission rate of 3.5% after home phototherapy, with evidence quality rated low for all outcomes [claim 40: unconfirmed], and a 2024 meta-analysis of 9 studies and 998 patients [claim 39: unconfirmed].
What would settle it: Bench irradiance mapping on a neonatal phantom across the covered surface and across the wear cycle, against the guideline target — and a read of the pilot's full text for whether the escalations tracked delivered dose or infant physiology, which is the question that decides whether this is a format problem or a patient-selection one.
If it's false: Under-dosing is a missed-escalation path, which is patient harm rather than a performance shortfall — Factor 6 goes to its floor, and the indication's exclusion of sicker infants stops being sufficient protection.
Would have to be true: A 510(k) against an LBI predicate would have to be achievable on bench photometric evidence with no clinical study — and the build would have to stay strictly inside a treatment-only claim.
Where it stands: The classification is verified — Class II, 21 CFR 880.5700, product code LBI, via 510(k) [claim 1: verified] — and predicate availability is explicitly not the constraint [claim 2: verified]. Whether clearance comes on bench data alone is still unchecked [claim 30: unconfirmed], and the pool to read it from is now named: 10 clearances in ten years, two of them wearable garments cleared in 2025-2026 [claim 47: unconfirmed]. The carried measurement extension remains classified but not live [claim 31: verified], with its transcutaneous route unidentified [claim 32: unconfirmed].
What would settle it: Read the 510(k) summaries for K251308 and K243372 for the test batteries actually accepted — the most useful unspent hour on this deck, and now precisely addressed; IEC 60601-2-50 for photometric and thermal requirements; FDA's home-use device human-factors guidance for the usability file.
If it's false: Clinical data lands in the critical path and the file's unbenchmarked estimate [claim 16: unconfirmed] is wrong by a multiple. If instead the measurement claim is added to win the monitoring argument, the whole Factor 1 verification lapses and the submission moves to a different review panel.
Would have to be true: An onboard radiometer would have to measure what actually reaches the skin — in a garment that shifts against a moving newborn — accurately enough that the number can be put in front of a clinician and a payer.
Where it stands: Unbacked [claim 27: unconfirmed], and the Verifier's disposition is that no connector in this repo reaches a bench sensor-agreement result. It is the only remaining differentiator: the cleared analog's indication already covers this candidate's whole target population and more, caregiver-administered, at home, claiming no dosimetry [claim 2: verified]. Two new facts raise the stakes. The pilot evidence says wearable delivery sometimes fails to control bilirubin [claim 41: unconfirmed], which is precisely what a trustworthy dose record would let a clinician catch early — the strongest argument for the feature anyone has yet had. And an incumbent blanket was recalled twice for discoloration/degradation of the fiberoptic bundle at the pad-to-box connection [claim 46: unconfirmed], which is a durability warning for any garment carrying optics and a sensor through repeated flexing and reprocessing.
What would settle it: A bench rig comparing onboard sensor output against a reference photometer at the skin plane, across the garment's real range of displacement and across phantoms at both ends of the size range [claim 35: unconfirmed], with a stated tolerance derived from the guideline target.
If it's false: The dose record is decorative, and what remains is a commodity LED wrap competing against cleared incumbents with broader labels [claim 2: verified], [claim 47: unconfirmed] on price alone.
Would have to be true: Irradiance mapping, sensor agreement across the size range, thermal safety and battery endurance would all have to be demonstrated on a phantom before any newborn is involved.
Where it stands: All the bench conditions are this deck's own write-back and all came back unconfirmed [claim 26: unconfirmed], [claim 27: unconfirmed], [claim 35: unconfirmed] — engineering claims about unbuilt hardware that no connector here reaches. Thermal, battery and home human factors have no claim at all [no claim]. The order of work is set by [claim 41: unconfirmed]: if a wearable can fail to control bilirubin, delivered-dose mapping is not one bench item among four, it is the one that determines whether the product is safe to put in a home.
What would settle it: IEC 60601-2-50 defines most of this test battery; the AAP guideline defines the target. Nothing here needs a human subject, which is the point of the slide.
If it's false: Any move toward clinical use is premature. Note the boundary: what happens after bench evidence is a Tier 3 decision, and no agent in this repo greenlights it.
Would have to be true: A rental unit would have to survive being cleaned between patients, delivered by van, operated by an exhausted parent at 3am, and returned.
Where it stands: Reprocessing-versus-single-patient-use economics are on record and unchecked [claim 29: unconfirmed]. The recall record gives that condition a concrete shape it did not have: the incumbent blanket's two recalls were for degradation at a flex-and-connect point [claim 46: unconfirmed] — the failure this product's form factor inherits and then compounds, because a garment flexes more than a pad and carries a sensor as well as emitters. Home human factors, thermal behaviour and battery endurance remain absent from the file entirely [no claim].
What would settle it: The predicates' 510(k) summaries for their reprocessing and usability approach [claim 47: unconfirmed]; a materials and cleaning-validation pass; a use-related risk analysis against FDA's home-use guidance.
If it's false: Unit cost per episode rises until the allowable cannot cover it, which lands straight on the supplier-margin condition [claim 24: unconfirmed] — now sharper, since the first observed allowable is $33.71 [claim 44: unconfirmed].
Would have to be true: The wearable-garment-plus-integrated-dosimetry combination would have to be buildable without infringing live art — and someone would have to have looked.
Where it stands: Nobody has looked. The claim is on record, names no patent number, and was never attempted [claim 28: unconfirmed] — unexamined, not clear. The field has become more crowded in exactly the relevant form factor since the last build, with two wearable garment clearances in 2025-2026 [claim 47: unconfirmed], which raises rather than lowers the prior that live art exists.
What would settle it: Credentials supplied, then a landscape read on wearable phototherapy enclosures and on radiometric dose logging, starting from the assignees the file names and the BiliBlanket fiberoptic lineage, with expiry dates on anything live. A Tier 3 legal opinion settles it; a connector-level read is enough to score the factor.
If it's false: Automatic kill under the Stage 5 gate — the only condition on any of these three decks with that property. An absence of found patents is not freedom to operate, and an unattempted claim is not even an absence of found patents.
Every condition above with nothing verified behind it.
The pattern is worth stating and the rebuild did not change it: all three verified claims in this file concern permission — the launch classification [claim 1: verified], the predicate pool and its labels [claim 2: verified], and the classification of a route the candidate is not currently taking [claim 31: verified]. None is engineering evidence. Fourteen new claims arrived since the last build, several of them squarely engineering-relevant, and every one landed unconfirmed. A reader skimming for green ticks will mistake permission for capability.
If only one thing from this chair could be checked: whether an onboard radiometer can measure delivered dose at the skin, in a garment that moves, within a defensible tolerance [claim 27: unconfirmed].
It held through the last verification and the new evidence makes it more load-bearing, not less, and for two independent reasons. Commercially it is the only differentiator left: the cleared analog's indication covers hospital and home use, caregiver-administered, across a wider population than this candidate targets, with no gestational-age floor and no bilirubin-band ceiling, and claims no dose record [claim 2: verified] — and there are now two cleared wearable garments in the class [claim 47: unconfirmed]. Clinically it has stopped being a documentation feature and become a safety instrument: a wearable device has been observed failing to control bilirubin in 2 of 12 infants [claim 41: unconfirmed], and a reliable delivered-dose reading is how a clinician would distinguish "this baby needs escalation" from "this wrap is not delivering".
If it holds, the dose record can carry the moat [claim 12: unconfirmed], answer the under-dosing risk [claim 14: unconfirmed], and potentially satisfy a descriptor that names a photometer [claim 4: unconfirmed]. If it does not, what remains is a commodity LED wrap in a class that already contains two cleared wearables. It is also cheap to settle — a bench rig, no subjects, no regulator, no payer.
Naming it is not a recommendation, a gate, or a kill.