idea-004 · human · US
For use in the home under prescription to deliver phototherapy to newborns of at least 35 weeks gestational age with uncomplicated indirect hyperbilirubinemia whose total serum bilirubin is above the phototherapy threshold but below the escalation-of-care threshold, as an alternative to inpatient readmission for phototherapy.
Finished enough to read, and waiting for you. No department has been asked. 6 of 76 claims verified. Nobody has yet written down what it rests on.
A wearable light-therapy wrap for newborns with jaundice, so the baby can be treated at home instead of staying in or returning to hospital. Jaundiced newborns are treated by lying under blue light, which breaks down the bilirubin causing it; today that usually means an inpatient stay whose only purpose is holding the baby under a lamp. This is a soft wrap with LEDs built in, meant to be rented to the family by a medical-equipment supplier and worn for most of the day. Its distinguishing feature is a built-in light sensor that records how much light actually reached the skin and for how long, so the prescribing clinician can see whether the baby really got the treatment. It deliberately does not measure bilirubin — adding that would put it on a slower, more expensive regulatory route. One thing about who buys it is newly open: if the insurers that pay for newborn home treatment pay it as a daily home-care service rather than as rented equipment, the buyer is a home-health agency, not an equipment supplier, and the device is a cost inside someone else's day rate.
In the Wrap's one-paragraph version, written for someone who has not opened the file. Domain: neonatal-hyperbilirubinemia-home-phototherapy
There are two cheques in this chain, written by different people for different things: the party that buys the device is not the party that pays for the therapy, and the second one sets the ceiling on the first. What changed since the last build is that the code and its Medicare rate are now on the record. What did not change is that neither of them belongs to the payer this population actually has.
adoption returns unverifiable on that line and the domain dossier
records that no source located names the actual US purchaser of a home phototherapy unit. The
first link in the chain is still an assumption.
hcpcs S9098) bundles equipment rental with nursing services
and blood draw. If the payers that matter pay home phototherapy under S9098 rather than E0202
[assumption A1], the device is not paid for as a device at all: it is a cost inside a home
health provider's per-diem, and the buyer becomes that provider rather than a DME supplier.
What this idea rests on that is not yet known, in the order this idea's own decision queue puts it: worst if false first, cheapest to settle first inside that. The first row is open; open any other for what would settle it.
This idea is unassessed, so nobody has yet named the few things it rests on. Until then, the gaps below are the rubric factors that are unscored or carry an open question. /assess idea-004 replaces this list with the real one.
Why it scored what it did: Claims "Class II (FDA device class II) under 21 CFR (Title 21 of the Code of Federal Regulations) 880.5700, product code LBI, 510(k) (premarket notification)" and "predicate availability is not the constraint (SnugLit K251308, 2025, wearable)". intended_use: is present and deliberately scoped to treatment-only, 35+ weeks, below…the rest of this reasoning
escalation threshold — which is what keeps the launch route on LBI. Not 5 because the photometer/dosimetry function invites special-controls scrutiny on photometric accuracy, and the carried bilirubinometer route in device_class is a separate submission the moment any measurement claim is made.
What would settle it: openFDA classification and 510(k) records for 880.5700/LBI class and predicate pool (free API lookup); the K251308 clearance summary PDF's indications-for-use, read directly (free, human read).
Why it scored what it did: Claims "HCPCS (Healthcare Common Procedure Coding System) E0202 exists, descriptor 'Phototherapy (bilirubin) light with photometer'", "the descriptor requires both light and photometer", "E0202 carries a published DMEPOS allowable" (amount unknown, not guessed), and the structural…the rest of this reasoning
note that newborns are not Medicare beneficiaries. A named DME code with an existing rental supply channel is materially better than no code. Held to 3, not higher, because: the allowable is unknown, no coverage policy has been checked, and the payers that actually matter here (Medicaid, ~40%+ of US births, and commercial) are outside what this repo's Medicare-anchored connectors can reach. Code existence alone is a code, not a pathway.
What would settle it: CMS (Centers for Medicare & Medicaid Services) HCPCS/DMEPOS/coverage files for the descriptor, allowable and Medicare policy (free, once placed in data/); the decisive sources are state Medicaid DME fee schedules and a commercial payer medical policy (e.g. Aetna/Cigna), both…the rest of this gap analysis
requiring human attachment.
Why it scored what it did: Rests entirely on the four decomposed TAM (total addressable market) claims: ~3.6M US births, 8-12% phototherapy rate at 35+ weeks, 25-40% home-eligible share, revenue per episode pegged to the E0202 rental allowable. Multiplied through, the midpoints imply very…the rest of this reasoning
roughly 100-120k home episodes/yr at DME-rental economics — a real and well-defined population, but a modest revenue pool that is fully dependent on the unknown E0202 allowable. Score also carries the AAP-2022-thresholds claim, which the Generator correctly flags as cutting against TAM (raised initiation thresholds shrink the treated population). Every input is a Generator guess; component 2 is self-described as the least defensible number in the file.
What would settle it: NCHS/CDC WONDER natality data for births, the 2022 AAP guideline and a birth-cohort study (e.g. Kaiser NCC) for phototherapy incidence — all free but need human attachment; home-eligible share likely needs a licensed payer-claims report.
Why it scored what it did: The Claims section contains no IP (intellectual property) or freedom-to-operate claim of any kind. Nothing is scored here rather than guessing a number: FTO (freedom to operate) is one of the four gate-relevant factors and the only one whose adverse finding is an automatic kill,…the rest of this reasoning
so a placeholder score would be worse than a blank. This is a Generator gap to close before or during Stage 3, not a Verifier failure.
What would settle it: A PatentsView landscape search on wearable-phototherapy/onboard-dosimetry claims, scoped to Natus/GE/Thera B and the BiliBlanket lineage (free API, needs PATENTSVIEW_API_KEY); a Tier 3 legal FTO opinion is the eventual settling source.
Why it scored what it did: Claims "the defensible assets are the dosimetry record and the DME distribution relationship, not the LED array, which is a commodity" and the explicitly-labelled assumption that "the binding constraint on adoption is the service model,…the rest of this reasoning
not device performance". Both point the same way and both are unflattering: if the service model is the constraint, the moat is a logistics and payer-contracting operation that a well-capitalised DME incumbent already has. The closest analog (SnugLit, wearable, cleared 2025) is named in the file as not a differentiator. Scored 2 rather than 1 only because the dosimetry-record asset compounds if it becomes a billing or coverage condition.
What would settle it: The FTO/IP patent landscape above; a payer LCD (local coverage determination)/Medicaid policy requiring a documented irradiance/adherence record for payment (human-attached); and whether the E0202 descriptor's photometer requirement is binding.
Why it scored what it did: Claim "phototherapy efficacy itself is long established; the device-specific risk is under-dosing at home, since delivered irradiance falls with distance and reduced body-surface coverage, with no bedside nurse checking either". The…the rest of this reasoning
mechanism is not in question — this is the strongest factor in the file. The residual risk is a dose-delivery engineering problem with a named mitigation (onboard radiometer) and a bench-testable spec, not a biological unknown. Not 5 because under-dosing at home is a real patient-harm path (missed escalation) and the dosimetry accuracy spec is entirely unvalidated.
What would settle it: The 2022 AAP guideline's irradiance spec and IEC 60601-2-50 for the dose target and bench-test protocol (free/public standards); openFDA recalls for LBI's historical failure modes (free lookup).
Why it scored what it did: Claim "a 510(k) against an existing LBI predicate with bench photometric testing and no clinical study is plausibly a low-single-digit-million, 12-24 month path to first revenue" — self-described as unbenchmarked. Taken alone that supports…the rest of this reasoning
a 4. Pulled to 3 by the service-model assumption claim: if the real product is same-day delivery, in-home setup, parent training and prior-auth handling, then first revenue also requires a rental fleet and a logistics operation, which is working-capital-heavy and slow to scale regardless of how cheap the 510(k) is. Regulatory cost is low; go-to-market capital is not.
What would settle it: openFDA 510(k) decision dates and summaries for LBI to benchmark regulatory cost/time (free); DME operator unit economics need a licensed industry report or operator interview (human-attached, cost varies).
Why it scored what it did: Claim "the LBI class contains established incumbents (Natus, GE) plus at least one recent wearable entrant, and the home segment is served by DME suppliers renting fiberoptic blankets", plus the predicate_or_analog note that SnugLit…the rest of this reasoning
(K251308, 2025) is "the closest analog, not a differentiator". A crowded class with entrenched distribution and a 2025 wearable entrant already occupying the proposed form factor is a low score. Scored 2, not 1, because the file identifies no entrant combining wearable form factor with onboard billing-linked dosimetry.
What would settle it: openFDA clearances/recalls for LBI entrants (free); funding and traction data need a licensed source (PitchBook/Crunchbase/CB Insights, subscription) or a human market scan for DME rental-channel ownership.
Why it scored what it did: No claim in the file addresses portfolio fit, and the repo contains no portfolio thesis or capital-allocation stance to score against — knowledge-base/ has other candidates but no statement of what the portfolio is trying to be. Guessing…the rest of this reasoning
a number here would be inventing the yardstick and the measurement in one step. Descriptively: this reads as a cash-generative DME spoke (fast, cheap clearance, recurring rental revenue, low technical risk) rather than a hero bet, which is a fit judgement a human owns.
What would settle it: A Tier 3 human-supplied portfolio thesis / capital-allocation stance — no connector or lookup can supply this; cost is a strategy discussion, not a query.
Seven chairs, each answering from its own field. The verdict, and the first condition it set.
Not convened. No department has stated a position on this idea yet — a normal state, not a defect. /round-table idea-004 asks all seven.
Only verified counts toward a score. A low count is usually a statement about the tooling — an absent data file, an unset key — rather than about the idea.
No claim on this idea has been refuted.
A human reads the state Medicaid DME fee schedules and home-phototherapy policies of the largest US birth states, plus one commercial medical policy, and records which code — E0202 or S9098 — each pays newborn home phototherapy under. Selected from the Reimbursement row's "What would settle it" cell, whose own words are "the decisive sources are state Medicaid DME fee schedules and a commercial payer medical policy (e.g. Aetna/Cigna), both requiring human attachment". The other leg of that cell — the CMS HCPCS/DMEPOS/coverage files — has now been run: that is what moved claims 3, 4 and 5 to verified, so it is no longer the cheapest thing left in the row. The cost is the one the viability deck gives: "a document read, by a human, of fee schedules that exist". It is the input the viability deck's load-bearing condition turns on claim 68unverified, and the same read bears on whether E0202 spend already exists for a home blanket to be substituted claim 76unverified and whether payers edit an E0202 claim for a missing photometer claim 73unverified. No connector reaches state Medicaid or commercial payer policy, so this is human work by construction, not a tooling gap anyone can close. With no assessment there is one next test, not two: no decision-mover exists to set beside it. This is a selection from an existing gap analysis, not a new judgement, and not a view about the idea.
Only verified counts toward a score. A low count is usually a statement about the tooling (an absent data file, an unset key) rather than about the idea.
Re-evaluated 2026-08-25 against the Verified Composite above (superseding the 2026-08-23 evaluation; the verdict is unchanged, but one statement below was factually wrong and the claim ledger it rested on was stale), using verified claims only. The Stage 2 draft table was not consulted. Unchecked boxes below mean the condition did not fire — in this file, in every case, because the data the condition reads does not exist yet. A condition that cannot be evaluated is not a condition that passed.
unconfirmed (market is unverifiable by design). An undefined operand is
not a low one, and the floor was not applied.
unverified and has never been attempted, because
PATENTSVIEW_API_KEY is unset. The factor is blank either way, but the blocker is an
environment task, not a Generator writing task — a materially different work item, and
the reason the correction is worth making rather than quietly fixing. **Zero patents were
searched.** FTO is the rubric's only automatic kill regardless of other scores, which makes
an unexamined field the most consequential hole here — an absence of found patents is not
freedom to operate.
Every one of the three kill conditions is written over a core factor — Regulatory, Reimbursement, TAM, FTO. Of those four, exactly one (Regulatory, 4/5) has a verified claim behind it, and it sits above its floor. The other three — the three that actually decide this gate — are blank. So the gate is not reporting that this candidate survived scrutiny; it is reporting that the scrutiny the gate performs has nothing to read. Recording that distinction is the whole point of separating queued-research from gated-out: "we don't know yet" is not "the answer is no."
hcpcs (does E0202 exist), payment (what it pays) and coverage (is there a policy) each returned a missing-data/-cache unconfirmed. Nothing was searched, so even the weak "a code exists" position has no verified basis.
procedures connector is both missing its cache and structurally inapplicable here, since newborns have no Medicare exposure. A null from it would have been a wrong-denominator artifact, not a small market.
unconfirmed,
13 unverified, 0 refuted), replacing the stale "16 lines / 14 unresolved". The verdict
and every factor score are unchanged; what changed is that the file no longer under-reports
how much of itself is unresolved.
status: set to queued-research; stage: stays verified — the
Verifier attempted every claim, so the pass is complete and must not be regressed.
wrapped is not an approval — it means a human can now read this.
No human review is recorded. Shortlisting, legal and clinical sign-off and capital allocation are Tier 3: no agent in this repo makes or simulates one. The packet is wrapped, so it can be read now.
Runs only after a human shortlists the idea.
No loop record: nobody has written down what this idea rests on or convened a department on it. That is a normal state, not a defect. /front-end-loop idea-004 opens one.
Nobody has yet written down what this idea rests on and what it would cost to be wrong. Unassessed is a state, not a low score. /assess idea-004 adds it.
Click any step to see what it holds for this idea. donecurrentloopinghuman only
## Wrap section, as writtenThe packet a human reads. See docs/wrap-spec.md. This section says nothing about whether the idea is good — it says the file is finished enough to be read, and names what to do next. Filling it in advances no candidate and approves nothing.
The one-paragraph version. A wearable light-therapy wrap for newborns with jaundice, so the baby can be treated at home instead of staying in or returning to hospital. Jaundiced newborns are treated by lying under blue light, which breaks down the bilirubin causing it; today that usually means an inpatient stay whose only purpose is holding the baby under a lamp. This is a soft wrap with LEDs built in, meant to be rented to the family by a medical-equipment supplier and worn for most of the day. Its distinguishing feature is a built-in light sensor that records how much light actually reached the skin and for how long, so the prescribing clinician can see whether the baby really got the treatment. It deliberately does not measure bilirubin — adding that would put it on a slower, more expensive regulatory route. One thing about who buys it is newly open: if the insurers that pay for newborn home treatment pay it as a daily home-care service rather than as rented equipment, the buyer is a home-health agency, not an equipment supplier, and the device is a cost inside someone else's day rate.
What would have to be true, quoted from each deck's load-bearing condition:
No assessment exists for this candidate (wrap_status.py: "assessment: none — unassessed, not a zero"), so there is no decision queue and no top-two pillars to set beside these three lines. That is a state, not a zero, and it is the thing most worth adding before a human reads this.
Cheapest next test: A human reads the state Medicaid DME fee schedules and home-phototherapy policies of the largest US birth states, plus one commercial medical policy, and records which code — E0202 or S9098 — each pays newborn home phototherapy under. Selected from the Reimbursement row's "What would settle it" cell, whose own words are "the decisive sources are state Medicaid DME fee schedules and a commercial payer medical policy (e.g. Aetna/Cigna), both requiring human attachment". The other leg of that cell — the CMS HCPCS/DMEPOS/coverage files — has now been run: that is what moved claims 3, 4 and 5 to verified, so it is no longer the cheapest thing left in the row. The cost is the one the viability deck gives: "a document read, by a human, of fee schedules that exist". It is the input the viability deck's load-bearing condition turns on claim 68unverified, and the same read bears on whether E0202 spend already exists for a home blanket to be substituted claim 76unverified and whether payers edit an E0202 claim for a missing photometer claim 73unverified. No connector reaches state Medicaid or commercial payer policy, so this is human work by construction, not a tooling gap anyone can close. With no assessment there is one next test, not two: no decision-mover exists to set beside it. This is a selection from an existing gap analysis, not a new judgement, and not a view about the idea.
Where the evidence actually stands. From wrap_status.py: 6 of 76 claims verified, 61 unconfirmed, 9 unverified, none refuted; 7 of 9 factors scored and 3 of 4 core factors — read that carefully, because those two numbers count the Stage 2 draft scores table, not verified evidence. All 6 verified claims are unstamped by evidence class. The 9 unverified claims (68-76) were appended today by the business case, the decks and the payer messages and have never been attempted by a Verifier pass — that is a pass not yet run, not a finding. The Verified Composite block was last recomputed on 2026-08-29 and predates claims 3, 4 and 5 moving to verified; re-deriving it is Scorer work at /rescore, and this wrap reports no composite rather than quote a stale one.
The six verified claims concern permission and Medicare coding; none concerns capability or demand. They are the launch classification claim 1verified, the predicate pool and its labels claim 2verified, the classification of a measurement route the candidate is not currently taking claim 31verified, and — new today — that E0202 exists as "Phototherapy (bilirubin) light with photometer" claim 3verified, that its descriptor text names a photometer claim 4verified, and that Medicare publishes an allowable for it, on one monthly capped-rental line only, non-rural $75.86 (AZ) to $89.25 (AL) claim 5verified. Read the new three with their own limits: newborns are not Medicare beneficiaries claim 6unconfirmed, so the rate is paid by a payer with no exposure to this population; claim 4verified verifies descriptor wording, not that the photometer is a payment condition claim 73unverified; and the same Level II file carries S9098, a per-diem home phototherapy service code claim 4verified, which is why the viability deck's load-bearing condition moved upstream of supplier margin. A reader skimming for green ticks will mistake a Medicare code for a pathway.
Blocked on tooling or on a pass not yet run, not on a finding — none of these is evidence that something does not exist:
coverage E0202 now runs, and found no policy naming E0202 — but the cached file
is the Medicare Coverage Database's Billing & Coding Articles export only, so LCDs and NCDs
proper were not searched, and the connector tests whether a policy names a code, never
whether it is covered (both limits recorded in claim 5verified). It is also Medicare's database.
drg and apc returning missing-cache
unconfirmed — no cache file for either is in data/.
patent connector ran with PATENTSVIEW_API_KEY unset claim 28unconfirmed; no patent
search has ever been run for this candidate. The connector now queries EPO (European Patent Office) OPS, and
EPO_OPS_KEY / EPO_OPS_SECRET are set in this session's environment, so the blocker is now
an unrun Verifier pass rather than a missing key. Even run, a sweep that finds nothing is not
freedom to operate, and the connector reads no legal status. FTO is unexamined, not clear.
data/pfs_rvu.csv is cached, so the business case's no-RVU-row assumption
claim 70unverified is now checkable by a connector; it simply has not been checked.
market / adoption are unverifiable by design. The TAM
inputs, clinician substitution claim 19unconfirmed and the one Medicaid rate on record —
Louisiana, $33.71 claim 44unconfirmed, on a billing unit nobody has established
claim 51unconfirmed — stay where they are until a human attaches a source.
Three things the research lane added that a Stage 6 reader should have, none of which moved a tag. The market is better measured: US births are a sourced figure rather than an estimate claim 36unconfirmed, with two US birth cohorts behind the eligible band claim 37unconfirmed, claim 38unconfirmed. The commercial risk is sharper: an implementation study of home phototherapy names lack of reimbursement options as one of the two main barriers to the service existing at all claim 42unconfirmed. The device risk has its first evidence: a first-in-human pilot of a wearable phototherapy device in 12 neonates completed treatment within 48 hours in 83%, with 2 of 12 switched to conventional intensive phototherapy because bilirubin kept rising despite wearable treatment claim 41unconfirmed — a small study of a different device, and the exact failure mode the file names as this candidate's device-specific risk claim 14unconfirmed.
One competitive fact belongs in the same breath. The class contains two cleared wrap/wearable garments from 2025-2026, BiliWrap and SnugLit, within 10 LBI clearances over ten years claim 47unconfirmed, and the cleared analog's indication already covers this candidate's whole target population and more, caregiver-administered, at home claim 2verified. The narrowed 35-week in-band indication is therefore not a commercial differentiator, and the differentiation case rests on the dose record — which is unbuilt, unmeasured, and the feasibility deck's load-bearing condition claim 27unconfirmed.
| Factor | Score | Why | What would settle it |
|---|---|---|---|
| Regulatory pathway | 4 | Claims "Class II under 21 CFR 880.5700, product code LBI, 510(k)" and "predicate availability is not the constraint (SnugLit K251308, 2025, wearable)". intended_use: is present and deliberately scoped to treatment-only, 35+ weeks, below…the rest of this reasoningescalation threshold — which is what keeps the launch route on LBI. Not 5 because the photometer/dosimetry function invites special-controls scrutiny on photometric accuracy, and the carried bilirubinometer route indevice_class is a separate submission the moment any measurement claim is made. | openFDA classification and 510(k) records for 880.5700/LBI class and predicate pool (free API lookup); the K251308 clearance summary PDF's indications-for-use, read directly (free, human read). |
| Reimbursement pathway | 3 | Claims "HCPCS E0202 exists, descriptor 'Phototherapy (bilirubin) light with photometer'", "the descriptor requires both light and photometer", "E0202 carries a published DMEPOS allowable" (amount unknown, not guessed), and the structural…the rest of this reasoningnote that newborns are not Medicare beneficiaries. A named DME code with an existing rental supply channel is materially better than no code. Held to 3, not higher, because: the allowable is unknown, no coverage policy has been checked, and the payers that actually matter here (Medicaid, ~40%+ of US births, and commercial) are outside what this repo's Medicare-anchored connectors can reach. Code existence alone is a code, not a pathway. | CMS HCPCS/DMEPOS/coverage files for the descriptor, allowable and Medicare policy (free, once placed in data/); the decisive sources are state Medicaid DME fee schedules and a commercial payer medical policy (e.g. Aetna/Cigna), both…the rest of this gap analysisrequiring human attachment. |
| TAM | 3 | Rests entirely on the four decomposed TAM claims: ~3.6M US births, 8-12% phototherapy rate at 35+ weeks, 25-40% home-eligible share, revenue per episode pegged to the E0202 rental allowable. Multiplied through, the midpoints imply very…the rest of this reasoningroughly 100-120k home episodes/yr at DME-rental economics — a real and well-defined population, but a modest revenue pool that is fully dependent on the unknown E0202 allowable. Score also carries the AAP-2022-thresholds claim, which the Generator correctly flags as cutting against TAM (raised initiation thresholds shrink the treated population). Every input is a Generator guess; component 2 is self-described as the least defensible number in the file. | NCHS/CDC WONDER natality data for births, the 2022 AAP guideline and a birth-cohort study (e.g. Kaiser NCC) for phototherapy incidence — all free but need human attachment; home-eligible share likely needs a licensed payer-claims report. |
| FTO / IP | — | The Claims section contains no IP or freedom-to-operate claim of any kind. Nothing is scored here rather than guessing a number: FTO is one of the four gate-relevant factors and the only one whose adverse finding is an automatic kill,…the rest of this reasoningso a placeholder score would be worse than a blank. This is a Generator gap to close before or during Stage 3, not a Verifier failure. | A PatentsView landscape search on wearable-phototherapy/onboard-dosimetry claims, scoped to Natus/GE/Thera B and the BiliBlanket lineage (free API, needs PATENTSVIEW_API_KEY); a Tier 3 legal FTO opinion is the eventual settling source. |
| Moat / defensibility | 2 | Claims "the defensible assets are the dosimetry record and the DME distribution relationship, not the LED array, which is a commodity" and the explicitly-labelled assumption that "the binding constraint on adoption is the service model,…the rest of this reasoningnot device performance". Both point the same way and both are unflattering: if the service model is the constraint, the moat is a logistics and payer-contracting operation that a well-capitalised DME incumbent already has. The closest analog (SnugLit, wearable, cleared 2025) is named in the file as not a differentiator. Scored 2 rather than 1 only because the dosimetry-record asset compounds if it becomes a billing or coverage condition. | The FTO/IP patent landscape above; a payer LCD/Medicaid policy requiring a documented irradiance/adherence record for payment (human-attached); and whether the E0202 descriptor's photometer requirement is binding. |
| Mechanism & clinical risk | 4 | Claim "phototherapy efficacy itself is long established; the device-specific risk is under-dosing at home, since delivered irradiance falls with distance and reduced body-surface coverage, with no bedside nurse checking either". The…the rest of this reasoningmechanism is not in question — this is the strongest factor in the file. The residual risk is a dose-delivery engineering problem with a named mitigation (onboard radiometer) and a bench-testable spec, not a biological unknown. Not 5 because under-dosing at home is a real patient-harm path (missed escalation) and the dosimetry accuracy spec is entirely unvalidated. | The 2022 AAP guideline's irradiance spec and IEC 60601-2-50 for the dose target and bench-test protocol (free/public standards); openFDA recalls for LBI's historical failure modes (free lookup). |
| Capital intensity & time-to-revenue | 3 | Claim "a 510(k) against an existing LBI predicate with bench photometric testing and no clinical study is plausibly a low-single-digit-million, 12-24 month path to first revenue" — self-described as unbenchmarked. Taken alone that supports…the rest of this reasoninga 4. Pulled to 3 by the service-model assumption claim: if the real product is same-day delivery, in-home setup, parent training and prior-auth handling, then first revenue also requires a rental fleet and a logistics operation, which is working-capital-heavy and slow to scale regardless of how cheap the 510(k) is. Regulatory cost is low; go-to-market capital is not. | openFDA 510(k) decision dates and summaries for LBI to benchmark regulatory cost/time (free); DME operator unit economics need a licensed industry report or operator interview (human-attached, cost varies). |
| Competitive intensity | 2 | Claim "the LBI class contains established incumbents (Natus, GE) plus at least one recent wearable entrant, and the home segment is served by DME suppliers renting fiberoptic blankets", plus the predicate_or_analog note that SnugLit…the rest of this reasoning(K251308, 2025) is "the closest analog, not a differentiator". A crowded class with entrenched distribution and a 2025 wearable entrant already occupying the proposed form factor is a low score. Scored 2, not 1, because the file identifies no entrant combining wearable form factor with onboard billing-linked dosimetry. | openFDA clearances/recalls for LBI entrants (free); funding and traction data need a licensed source (PitchBook/Crunchbase/CB Insights, subscription) or a human market scan for DME rental-channel ownership. |
| Strategic fit & portfolio balance | — | No claim in the file addresses portfolio fit, and the repo contains no portfolio thesis or capital-allocation stance to score against — knowledge-base/ has other candidates but no statement of what the portfolio is trying to be. Guessing…the rest of this reasoninga number here would be inventing the yardstick and the measurement in one step. Descriptively: this reads as a cash-generative DME spoke (fast, cheap clearance, recurring rental revenue, low technical risk) rather than a hero bet, which is a fit judgement a human owns. | A Tier 3 human-supplied portfolio thesis / capital-allocation stance — no connector or lookup can supply this; cost is a strategy discussion, not a query. |
All eight wrap criteria are met.
Source file: knowledge-base/candidates/idea-004.md
Nothing on this page is a recommendation. Shortlist review, legal and clinical sign-off and capital allocation are human-only decisions, and no agent in this repo may make or simulate one. This is a compilation of what the candidate file says.
_shared/plain-language.md