idea-004 · 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 turn a rental fleet, a van and a same-day delivery promise into a business.
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, and the tag is copied from that line and never adjusted here. A [no claim] marker means nothing in the file speaks to the condition at all. No number appears here that is not in a claim, and there is no ask slide. See docs/deck-spec.md.
What changed since the last build — read this first. The Verifier's 2026-09-29 passes moved the three reimbursement legs this deck used to report as blocked on a missing cache: E0202 (a HCPCS code for durable medical equipment) exists claim 3verified, its descriptor names a photometer claim 4verified, and Medicare publishes an allowable for it — one line only, a monthly capped rental claim 5verified. That is real, and it is narrower than it looks: every verified leg is Medicare's, and newborns are not Medicare beneficiaries claim 6unconfirmed. The business case rebuilt the same day put four payer-side assumptions on record (claims 68-71), and one of them moves this deck's load-bearing condition upstream of where it sat; see slide 9. This deck put one new condition on record: whether the photometer is a payment condition claim 73unverified. Rebuilt from the file, not patched.
Would have to be true: A DME (durable medical equipment) supplier would have to rent this unit to a family and bill a state Medicaid program or a commercial plan under the equipment code E0202 — so that the company's customer is the supplier and the money comes out of the payer's DME benefit, not out of a home-health per-diem, a hospital budget or the family's pocket.
Where it stands: The supplier-as-customer premise is unconfirmed claim 22unconfirmed. The payers that matter are Medicaid and commercial because newborns are not Medicare beneficiaries claim 6unconfirmed, and whether Medicaid rather than commercial rates set the effective price is unconfirmed claim 63unconfirmed. The new question sits under all of this: the business case assumes the payers that matter may pay newborn home phototherapy predominantly under S9098 (a HCPCS temporary national code), a per-diem home phototherapy service code, rather than under E0202 claim 68unverified. S9098 exists in the Level II file as a separate per-diem service code claim 4verified. If the payers use it, the device is a cost inside a home-health provider's day rate and the customer is that provider. On the other side of the ledger, the avoided readmission may be paid to the hospital as a per-case DRG, so the saving accrues to the payer, not the hospital claim 71unverified — which rules the hospital out as a natural buyer.
What would settle it: One large-birth-state Medicaid DME and home-health fee schedule read by hand for which code carries home phototherapy; a named commercial plan's medical policy for the same; two or three DME operator interviews on who buys and who carries the fleet capital.
If it's false: If the payers pay S9098, the customer, the unit economics and the moat argument all change at once: a home-health provider buys a cost input, and the E0202 descriptor's photometer (slides 2 and 5) stops mattering. Factor 2 and Factor 5 both re-open.
Would have to be true: For the payers this population actually has, a code would have to exist, pay an amount on a billing unit that fits a 2-4 day episode, and be covered for this indication — three separate conditions, all three landing.
Where it stands: One verified for the wrong payer, two open.
What would settle it: Largest-birth-state Medicaid DME manuals and fee schedules for E0202's billing unit and coverage criteria, read by hand; the Louisiana schedule itself for its unit. No connector in this repo reaches a state Medicaid or commercial policy.
If it's false: Factor 2 is blank in the file, not low. A bad answer on payment unit or coverage is the half of the regulatory-and-reimbursement kill condition that remains unevaluated; the regulatory half is above floor, so the compound condition cannot fire, but the business can still fail without a gate saying so.
Would have to be true: Enough eligible home episodes a year, at enough revenue per episode, to be worth a rental fleet.
Where it stands:
What would settle it: NCHS natality by gestational age and source of payment (free, outside the connectors — a human attachment); a Medicaid claims extract for the home-eligible share; the state billing unit from slide 2.
If it's false: TAM (total addressable market) is one of the three hard kill conditions, so this is where a bad answer is fatal rather than discouraging. Nothing has fired: the components are unmeasured, not adverse.
Would have to be true: Clearance would have to be cheap and short, and a fleet would have to reach paying episodes — through supplier enrollment, state Medicaid enrollment, prior authorisation and the claims cycle — on a calendar someone can finance. No claim carries a cost or a calendar, so none is given here.
Where it stands: The regulatory half is the best-supported thing in the file: Class II (FDA device class II), LBI, 510(k) (premarket notification) claim 1verified, predicate availability not the constraint claim 2verified. Whether clearance comes on bench data alone is unconfirmed claim 30unconfirmed, and the time-and-cost estimate is unbenchmarked claim 16unconfirmed. The commercial half is all open: the lag to first paid claim claim 58unconfirmed, channel build in the largest birth states claim 59unconfirmed, DMEPOS (durable medical equipment, prosthetics, orthotics and supplies) and Medicaid enrollment cost claim 60unconfirmed, fleet payback claim 54unconfirmed, and the service-model assumption that logistics, not device performance, is the binding constraint claim 13unconfirmed.
What would settle it: A human reading the two 2025-2026 wearable 510(k) summaries for whether clinical data was required; a DME operator's unit economics and days-to-payment from an interview. No connector returns submission cost, review time or supplier economics.
If it's false: A cheap clearance in front of an expensive, slow logistics business — Factor 7's draft rationale already says as much, and the verified composite leaves Factor 7 blank.
Would have to be true: The payers that matter would have to refuse or recode an E0202 claim for a light supplied without a photometer — so the onboard dose record is a condition of being paid, not a feature a competitor can skip — and no one else could build the same record.
Where it stands: The descriptor wording is now verified claim 4verified, and the claim itself states the limit: it does not verify that the photometer binds for payment, which is DME MAC (durable medical equipment Medicare administrative contractor) and payer-edit territory. That payment half was never on record as its own claim; this build put it there claim 73unverified. Whether the record earns incremental payment is unconfirmed claim 57unconfirmed; the moat lever as a whole is unconfirmed claim 12unconfirmed; FTO (freedom to operate) was attempted but could not run and names no patent claim 28unconfirmed. And if slide 1's S9098 assumption holds claim 68unverified, the descriptor this whole argument leans on is not the one being billed.
What would settle it: A DME MAC supplier manual or PDAC (Pricing, Data Analysis and Coding contractor) coding guidance on E0202, and one state Medicaid DME edit for the code, both read by hand; a patent landscape search once a PATENTSVIEW_API_KEY is supplied.
If it's false: The dose record is promotional, not structural, and the device is a commodity LED wrap. Factor 5 is blank in the file; the file records adjacent verified evidence (the broad cleared analog) as adverse context that it expects to resolve low.
Would have to be true: A DME supplier stocking a home phototherapy fleet would have to choose a dosimetry-equipped wearable over the blanket already in its fleet and over the cleared wearables — for a reason that shows up in what it is paid.
Where it stands: The field is occupied and that is verified. Natus and GE are named incumbents and two wrap/wearable garments are cleared, SnugLit (decided 2026-01-15) and BiliWrap claim 2verified; SnugLit's cleared indication covers hospital and home use, caregiver-administered, across a wider population than this candidate targets, and claims no dose record claim 2verified. Ten LBI clearances in ten years claim 47unconfirmed. The supplier's choice is unconfirmed claim 34unconfirmed; whether SnugLit's maker holds the DME relationships is unconfirmed claim 25unconfirmed; the DME-channel half of the competitive claim is unconfirmed claim 15unconfirmed. Two adjacent signals: two historical recalls of the Natus blanket for fiberoptic-bundle degradation claim 46unconfirmed, and a completed US trial of a competing wearable with no results posted claim 48unconfirmed.
What would settle it: DME operator purchasing interviews; a licensed market scan for funding and traction, which no connector covers.
If it's false: Read inverted. The field is already known to be occupied; the risk is that suppliers are paid the same whatever they stock, in which case the cheapest unit wins.
Would have to be true: One afternoon of human reading would have to move the money question from assumption to fact — which code the payers that matter pay home phototherapy under, and on what billing unit.
Where it stands: The cache re-run the last build named as cheapest has happened, and it settled the Medicare legs claim 3verified, claim 4verified, claim 5verified. What it could not settle is the payer this population has claim 6unconfirmed. The state-level evidence in the file was found by hand claim 44unconfirmed, and the questions it opens are on record claim 51unconfirmed, claim 68unverified, claim 69unverified.
What would settle it: Two large-birth-state Medicaid fee schedules and DME/home-health manuals, read for E0202 versus S9098, billing unit, and any photometer or documentation edit. That one read touches slides 1, 2, 3 and 5.
If it's false: Nothing here is false — this slide is a work order.
Every condition above with nothing verified behind it — all seven. Slides 2, 4 and 6 each carry verified legs (code, Medicare fee, classification, predicate pool, the occupied field), but none of those legs is the condition itself: each verified fact is either Medicare's or a fact about permission and competitors.
Of the six rubric factors that live in this lens, the Verifier's week moved three claims under Reimbursement to verified. Reimbursement is still blank in the Verified Composite as it stands, which was last recomputed on 2026-08-29, before the tags moved; recomputing it is the Scorer's job, not this deck's. Whatever the recompute finds, every verified leg under it is Medicare's.
If only one thing from this chair could be checked: whether state Medicaid programs and commercial plans in the largest birth states pay newborn home phototherapy under the equipment code E0202 at all, rather than under the per-diem service code S9098 claim 68unverified.
This moved from the last build, and the reason is new claims rather than a new opinion. The last build named supplier margin under the E0202 allowable as load-bearing claim 24unconfirmed. That question presupposes the answer to this one. If the payers that matter pay S9098, there is no E0202 allowable for the margin to sit under, the photometer in the descriptor claim 4verified decides nothing, and the buyer is a home-health provider rather than a DME supplier. If they pay E0202, the margin question returns, now with a known billing-unit problem attached claim 69unverified. It is also the cheapest condition on the deck to settle: a document read, by a human, of fee schedules that exist.
Naming it is not a recommendation, a gate, or a kill.
_shared/plain-language.md