idea-010 · 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 assay scientists who have to get an accurate potassium out of a drop of blood that starts leaking intracellular potassium the moment the lancet lands.
How to read this: every slide is a condition, not a conclusion. A bracketed reference points into knowledge-base/candidates/idea-010.md — the number is the claim's position in its ## Claims list and the tag is copied from it, never adjusted here. [no claim] means nothing in the file speaks to the condition at all. See docs/deck-spec.md.
What the composite says about this deck: 60/100 is computed over Factor 1 alone (15% of weight, renormalised). Factor 6 — mechanism and clinical risk — is blank, meaning no verified claim, and it is the factor that decides whether this device can exist. The score does not read the risk; this deck does.
Would have to be true: A single-use strip would have to return a quantitative potassium and a quantitative creatinine from one fingerstick capillary drop, on a handheld meter, within a clinic visit, accurate enough that a prescriber will move a RAASi dose on it — and a third on-strip channel would have to decide, before display, whether the potassium result is fit to report.
Where it stands: The specification is in the frontmatter and it is specific about mechanism: an ionophore-membrane ISE potassium channel, an enzymatic creatinine channel, and a hemoglobin channel whose sole job is to detect hemolysis and suppress the potassium result rather than report a falsely elevated one [claim 22: unconfirmed]. The regulatory identity of each analyte channel is verified: potassium is 21 CFR 862.1600, Class II [claim 1: verified]; creatinine is 21 CFR 862.1225, Class II [claim 2: verified]; the codes are CEM [claim 3: verified] and CGL [claim 4: verified]. What no claim in the file carries is a number: no required bias, no imprecision limit, no drop volume, no time-to-result, no hematocrit range. The performance target is on record only through this deck [claim 39: unverified].
What would settle it: The special controls and any FDA-recognised consensus standards under the two regulations; CLIA acceptance limits for potassium and creatinine as the outer bound a waiver application would be judged against; CLSI EP09/EP35 for how the comparison is run. The predicate's own performance section — the FDA-posted summary for K070068 is already verified as readable [claim 5: verified] — gives the creatinine channel a benchmark that has cleared before.
If it's false: Without an acceptance criterion there is nothing to power the method-comparison study against, and Factor 7's evidence package cannot be scoped at all [claim 26: unconfirmed] — you cannot cost a study whose endpoint is undefined.
Would have to be true: Capillary fingerstick sampling would have to produce a potassium result that is either not materially elevated by hemolysis, or reliably identifiable as elevated by an on-strip hemoglobin channel at the free-hemoglobin concentration where the bias becomes clinically material — before the number reaches the screen.
Where it stands: The file states the physics plainly and against its own interest: capillary sampling lyses red cells, hemolysis releases intracellular potassium, and the resulting false elevation is the named reason a strip-based fingerstick potassium meter plausibly does not already exist [claim 22: unconfirmed]. It is a property of the sample, not an engineering defect awaiting effort — the hemolysis is generated at the finger, upstream of anything the strip can control. The magnitude the detection channel has to reach is nowhere in the file and is on record only through this deck [claim 40: unverified]. The Verified Composite records that no connector in this repo can reach any of it, because it is bench physics on a device that does not exist.
What would settle it: Bench work, and only bench work: an interference series establishing the free-hemoglobin concentration at which this strip's potassium channel exceeds the bias limit from Slide 1, and a detection study showing the hemoglobin channel resolves that concentration at fingerstick volume. Before spending on that, a human reads the published capillary-versus-venous literature at full text — the Stage 0 literature scan logged to knowledge-base/signals/log.md on 2026-08-25 names specific records including one on in-line hemolysis detection on a cleared analyzer, and it is a pointer carrying no confidence tag: it names papers to pull, it evidences nothing here, and it re-tags nothing.
If it's false: If the bias cannot be detected at fingerstick volume, the mitigation does not work and the device reports falsely elevated potassium — which in a titration decision means withholding a drug the patient should be on. That is the clinical-risk half of Factor 6 and it is worse than the commercial failure on Slide 4.
Would have to be true: FDA would have to accept a capillary-whole-blood strip inside two plasma- and serum-worded generic types, on two predicate arguments in one submission, alongside a CLIA waiver application built on untrained-operator studies.
Where it stands: One verified fact here cuts against the candidate and must be carried, not waved off. 21 CFR 862.1600's identification was read verbatim and names potassium measurement in "serum, plasma, and urine" — it does not name whole blood, capillary or fingerstick [claim 1: verified]. 862.1225 is narrower still, plasma and urine only, not even serum [claim 2: verified]. This device measures capillary whole blood. The counterweight is also verified and is the documented route through: K070068 sits under the plasma-worded 862.1225 with an IFU covering "capillary, venous, and arterial whole blood" for point-of-care use, and states the meter is "calibrated to provide plasma equivalent results to laboratory methods" [claim 5: verified] — which makes the same trick a requirement for the potassium channel, and that requirement is on record only through this deck [claim 41: unverified]. On the potassium predicate itself, the frontmatter's "no specific predicate identified" is overstated against the candidate's own interest: the audit enumerated the generic type across both product codes (CEM and MZV) over 30 years and found predicate candidates that exist — K200865 and K042270, both whole blood at point of care — with no device in this form factor among them [claim 7: unconfirmed]. That enumeration is a Stage 0 ScanResult with no confidence tag; whether either is an adequate predicate is an FDA substantial-equivalence judgement for a human. The combined Dual 510(k) and CLIA Waiver by Application pathway is structurally unconfirmable here — no connector reads FDA guidance [claim 8: unconfirmed] — and De Novo is entirely unsearched, so no De Novo absence may be inferred [claim 7: unconfirmed]. One further build item is on record through this deck: whether the hemolysis-suppression function itself must be evaluated in the marketing application rather than accepted as an internal safeguard [claim 43: unverified].
What would settle it: An FDA pre-submission on the capillary-whole-blood matrix and on the suppression function — that is the question a human asks FDA, not a database. FDA's De Novo listing into data/ to close the fork. The named Dual 510(k)/CLIA Waiver guidance PDF attached by a human. And the FDA-posted summaries for K200865 and K042270, already shown readable, read for how each handled its sample matrix. The recall records logged in knowledge-base/signals/log.md on 2026-08-25 are a pointer only — no confidence tag, no finding — but they are the entry a human should read first on the suppression-function question.
If it's false: If the matrix gap or the suppression function forces a De Novo, Factor 1's route and Factor 7's calendar move together — the file names that fork as the dominant term in both cost and schedule [claim 26: unconfirmed].
Would have to be true: The invalid-result rate in ordinary clinic hands — the fraction of attempts that end in a suppressed or unusable potassium — would have to be low enough that the clinic keeps reaching for the meter.
Where it stands: The file names this as the product's real performance metric, precisely because the mitigation suppresses rather than corrects: a strip that voids the sample produces no result rather than a wrong one, and the file states that a device voiding one draw in four is not clinically usable however accurate it is on the rest [claim 23: unconfirmed]. The upstream cause is the sampling physics itself [claim 22: unconfirmed], and the operator conditions that drive it — edematous, cool-peripheried, frequently-lanced patients, and staff tempted to milk the finger — are the desirability deck's Slide 3 [claim 32: unverified]. The Verified Composite records Factor 6 as blank and the Gate Check records that this risk contributed nothing to the Stage 5 result — it is invisible to every factor that was scored.
What would settle it: The Stage 7 study the file already specifies: a paired capillary-versus-venous method comparison across multiple ordinary operators with a deliberate lysis challenge, reporting the invalid/suppressed-result rate as a primary endpoint alongside bias — not a clean-sample accuracy study. A cheaper precursor exists and needs no strip: a human reading the published capillary-versus-venous literature at full text for reported hemolysis and rejection rates, and reading any cleared point-of-care potassium device's IFU for whether it excludes capillary specimens — an incumbent labelling capillary out is the market saying the problem is unsolved.
If it's false: The device is technically successful and commercially dead: staff learn that the meter fails often enough to be unreliable, order the venous draw first, and the workflow reverts. Every condition on the desirability deck falls with it, and so does the consumable annuity the viability case rests on.
Would have to be true: A bench rig would have to show detection of hemolysis at the bias- relevant threshold, and stable plasma-equivalent potassium across the hematocrit range these patients actually present with, before any human study is designed.
Where it stands: Neither exists. The detection threshold is on record only through this deck [claim 40: unverified], and the hematocrit condition likewise [claim 41: unverified] — and that one is not incidental: CKD and heart-failure populations are disproportionately anemic, so the calibration has to hold at the low end of hematocrit rather than at a healthy-volunteer mid-range. The file's own Factor 6 gap analysis names the definitive study, and states outright that no connector can settle any of this [claim 22: unconfirmed], [claim 23: unconfirmed].
What would settle it: Spiked-hemolysate interference series against a reference potassium method; hematocrit-adjusted whole-blood panels across the anemic range; then a small paired capillary/venous run with ordinary operators before anything is scaled. Tier 3 sits immediately downstream of this slide — no agent in this repo greenlights any study involving humans or animals, and the Diligence Lead drafts this list at Stage 7.
If it's false: The pivotal method-comparison study fails on invalid rate or on hematocrit bias rather than on the thing it was designed to measure, which is the most expensive possible place to learn it [claim 26: unconfirmed].
Would have to be true: A three-channel strip would have to be manufacturable at a yield and a cost that survive Slide 7 of the viability deck, hold calibration and its hemolysis threshold on a clinic shelf, and be run correctly by a medical assistant who is not a laboratory technologist.
Where it stands: The manufacturing leg is asserted as one of the moat assets — the strip yield curve, with the meter conceded as a commodity — on the Generator's judgement and not connector-checkable as written [claim 24: unconfirmed]. Shelf life and threshold stability under ordinary clinic storage are on record only through this deck [claim 42: unverified]. The operator leg is the CLIA waiver's whole subject: untrained-operator method-comparison studies at multiple sites [claim 26: unconfirmed], with the willingness of the clinic to carry the QC and competency burden an open assumption [claim 29: unconfirmed]. One genuinely favourable verified fact sits here: K091547 is a real cleared Nova Max Plus glucose/β-ketone meter taking a fingertip capillary drop and cleared for home lay use [claim 6: verified] — but read its scope exactly, because the file does: it is a format analog only, it is not a potassium or creatinine predicate, and it evidences the carried OTC/home-use route in device_class:, not the launch indication's predicate argument.
What would settle it: Accelerated and real-time stability protocols on the finished strip; a pilot line yield study, which is Tier 3 capital rather than a lookup. For the operator half, the CLIA waiver criteria in 42 CFR 493.15 — noting the file's audit found that lookup did not reproduce today owing to eCFR timeouts and a 503, an upstream availability failure and not a finding — plus the human-attached Dual 510(k)/CLIA Waiver guidance [claim 8: unconfirmed].
If it's false: Strip cost rises against an allowable nobody has retrieved [claim 14: unconfirmed], and the margin inequality that the viability deck turns on gets worse from the cost side while the payment side is still blank.
Would have to be true: An on-strip hemoglobin channel that gates a co-located potassium result would have to be buildable without infringing live art — and somebody would have to have looked.
Where it stands: Nobody has looked. No patent search was performed, no patent number is named, and PATENTSVIEW_API_KEY is unset, so even a named number would return unconfirmed [claim 25: unconfirmed]. The file is explicit that this is an unexamined field, not a clear one, and that the art is unusually dense — ISE potassium chemistry, enzymatic creatinine and glucose-strip electrochemistry are all mature and heavily filed by exactly the firms named as the competitive field [claim 27: unconfirmed]. The exposure concentrates on the differentiator rather than the hardware, since the differentiator is the moat [claim 24: unconfirmed] and the hemolysis-suppression logic is the only thing on this strip that is not commodity.
What would settle it: The file names the single highest-value query and this deck does not improve on it: art on hemoglobin or hemolysis detection used to flag or suppress a co-located electrolyte result. Patents Public Search / Espacenet across CPC G01N27/333, G01N33/49, G01N33/72, G01N27/327 and C12Q1/00, with assignee sweeps on Nova Biomedical, Abbott Point of Care, Roche, Siemens Healthineers and i-SENS; then connectors patent on each number once a key exists; then a Tier 3 legal FTO opinion at Stage 7.
If it's false: Automatic kill under the Stage 5 gate, and it lands on the one asset that distinguishes this device from the 34 cleared potassium systems the audit enumerated [claim 7: unconfirmed]. Stage 5 could not fire the condition here because an unexamined field is neither clear nor blocked — the exposure is real and unmeasured, not absent.
Every condition above with nothing verified behind it — six of the seven. Slide 3 is the exception and only in part: the pathway slide rests on four verified regulatory facts, and the sharpest of them, the "serum, plasma, and urine" identification text, is verified evidence pointing against the predicate argument rather than for it [claim 1: verified].
data/.
Two cautions on how this list may be read. Every enumeration and literature result cited in the "what would settle it" fields above is a Stage 0 ScanResult that carries no confidence tag and cannot acquire one — those are pointers to sources a human should pull, and converting one into evidence would be exactly the failure this pipeline exists to prevent. And five of the nine items above carry claim numbers only because this deck put them on record this morning as unverified: that makes them visible to the Verifier, it does not make them answered.
If only one thing from this chair could be checked: the invalid-result rate in ordinary clinic hands — what fraction of fingerstick attempts, by real clinic staff on real heart-failure and CKD patients, end with the potassium result suppressed rather than reported [claim 23: unconfirmed].
It is chosen over accuracy, and that is the whole point. Accuracy on clean samples is a solvable engineering problem with a known playbook; the file's own position is that the sampling physics is a property of the specimen rather than a defect awaiting effort [claim 22: unconfirmed], and the mitigation converts that physics into voided tests instead of wrong answers. So the number that decides whether this device is usable is not a bias figure — it is a rate of nothing-happening, which no clean-sample study reports, no connector in this repo can reach, and no factor in the rubric currently scores. Factor 6 is blank; the Gate Check states outright that the risk deciding this candidate contributed nothing to the Stage 5 result.
Naming it is not a recommendation, a gate, or a kill.