Hospital CKM | Continuous Potassium Monitoring Prototype

Why this product

ICU potassium is already managed as a continuous-risk variable. The data is still intermittent.

Potassium variability is associated with mortality in critical illness. Computerized potassium protocols reduce time outside target ranges and reduce hypo/hyperkalemia events. Hospital CKM turns that protocol logic into a sensor-backed surveillance product.

206,987Potassium measurements in a 10,451-patient ICU analysis linking potassium level, variability, and mortality.

23.3% → 19.9%In-hospital mortality change reported after computerized potassium control protocol implementation.

9.2% → 3.9%Time below 3.5 mmol/L after protocolized potassium control in critical-care data.

<2% CVReported potassium reproducibility benchmark for i-STAT point-of-care analyzer studies.

Clinical painICU teams respond to scheduled BMPs, blood gases, alarms, insulin shifts, CRRT prescriptions, diuretics, arrhythmias, and post-op physiology. Dangerous K+ movement can happen between samples.

Product thesisHospital CKM does not replace blood gas analyzers. It watches the interval between them and triggers earlier confirmation, protocol action, or line/circuit review.

Utilitarian ROIThe highest-value wedge is where the patient already has vascular access, dialysis/CRRT flow, or frequent lab sampling, and where a missed potassium trajectory changes the next hour of care.

ICU K+ variability Computerized protocol i-STAT benchmark

Product architecture

A closed disposable flow-cell with hospital-grade calibration controls.

The hospital version should not inherit wearable constraints. The bench system gets a larger cartridge, external calibration standards, measured fluid temperature, stronger diagnostics, and a bedrail reader. Battery backup, EHR export, active heating, and patient-circuit integration remain later engineering work.

01 / DISPOSABLE

Inline K+ flow-cell

Single-patient closed cassette with K1/K2/K3 valinomycin ISEs, blank control, dual references, temperature, pressure, bubble detection, and standard/calibration loop.

02 / REUSABLE

Bedside reader

Six guarded electrometer channels, 24-bit ADC, fluidic pump control, optical bubble checks, bridge-pressure acquisition, local display, isolated bench power, and raw data export.

03 / OUTPUT

Protocol trigger

Reports validated K+ trend, drift/QC state, rate of change, and confirmatory-test trigger. It never acts as autonomous potassium dosing.

Sample paths

Three deployment modes, one measurement stack.

Hospital CKM should be designed as a modular fluidic platform so the same sensor stack can be proven in the easiest matrix first, then advanced toward higher-acuity blood-contacting use.

PATH A

Dialysate / CRRT effluent

Lowest blood-contact risk. Tracks potassium exchange and prescription response around CRRT/dialysis workflows. Best first hospital feasibility path.

PATH B

Drawn-line micro-sample

Small periodic samples from arterial/venous access into a disposable measurement loop, then waste. Enables direct blood potassium trend without leaving sensor in the bloodstream.

PATH C

True inline blood

Highest clinical immediacy, highest burden: hemocompatibility, clotting, hemolysis, pressure drop, sterilization, line integration, and regulatory review.

Recommended startBuild Path A first: dialysate/CRRT/effluent and controlled blood-simulant rigs. It proves continuous electrolyte sensing without needing a first prototype in the blood path.

Clinical expansionMove to drawn-line micro-samples after bench performance and cassette sterility/containment are stable. True inline blood is the accuracy north star, not the first build.

In-line CRRT electrolytes CRRT electrolyte disturbances

Design-audited bench architecture

One first article, with the remaining decisions made explicit.

Potassium ISE chemistry is already used in clinical electrolyte and blood-gas analyzers. That precedent supports feasibility, not this device's accuracy. Hospital CKM still has to prove its exact sensor carrier, reference system, hydraulic refresh, guarded electronics, calibration, and fail-closed behavior.

$3.3K-$7.3KArithmetic sum of the detailed planning BOM before shipping, tax, and the integrated sensor-carrier RFQ. Existing inventory should reduce remaining spend.

≥60 secMinimum three-volume refresh for the current 0.61 mL cell at the flow sensor's 2 mL/min full-scale rate; tubing adds more delay.

2-9 mmol/LBench validation span covering hypokalemia, normal range, hyperkalemia, dialysis/CRRT transitions, and failure states.

Comparator firstResearch error targets do not become claims until matrix-matched standards and an independent analyzer establish bias.

Bench freezeKeep ordered DRP-110-class single-cell electrodes for coating and channel screening. Before machining the integrated cover, choose a measured multi-lane carrier: Metrohm DRP-8X110-U20 with CAC8X interface or a custom six-lane substrate. A DRP-110 has one 4 mm working electrode; it is not the six-pad carrier shown in the old render.

Reference freezeUse two independent double-junction Ag/AgCl references with a potassium-free outer bridge such as 10% KNO3. A single-junction KCl reference can leak the analyte into this low-volume cell and create false potassium drift.

Electronics freezeUse the two documented ADA4530-1ARZ parts for early channel bring-up, then standardize all six high-impedance nodes on one amplifier family before the guarded PCB is fabricated. ADS124S08 digitizes the buffered potentials and analog diagnostics; TMP117 and SLF3S flow are separate digital sensors.

Pass conditionAcross repeated 2.0, 3.0, 4.0, 5.5, 7.0, and 9.0 mmol/L steps, K1/K2/K3 agree, the matched blank remains quiet, dual references remain stable, at least three measured cell volumes have passed, and bubbles, stale flow, pressure faults, or reference movement suppress output.

TI ADS124S08 ADI ADA4530-1 TI TMP117 Metrohm 8X array Double-junction reference K+ ionophore cocktail In-line dialysis electrolyte monitoring

Build-readiness audit

Reuse what is already here. Buy only the gaps that block valid data.

The master inventory documents several reusable parts, but it does not prove every Hospital CKM item is physically in hand. Reconcile each packing slip before placing duplicate orders.

Documented inventory to verify and reuse

TMP117Adafruit breakout plus a bare sensor are documented as received. Use the breakout for first data and thermally couple it to the fluid bay.

ESP32-S3One ESP32-S3-DevKitC-1-N8R8 is documented as received. It is sufficient for USB data capture, pump control, and the bench display.

ADA4530-1Two ADA4530-1ARZ electrometer amplifiers are documented as received. Use them for leakage/noise bring-up; an integrated six-node build needs four more if this family wins the freeze.

PEDOT:PSSOssila PH1000 is documented as received but was left unrefrigerated. Use it only after film, conductivity, adhesion, and drift checks; replace it if those fail.

350 mAh LiPoTwo cells are documented as received. They can isolate a debug run from mains noise, but they are not a medical-grade reader power system.

Do not start integrated assembly without these closures

Sensor carrierMeasure and choose DRP-8X110-U20 + CAC8X or a custom six-lane carrier. Existing DRP-110 strips remain useful for chemistry screening.

ReferencesTwo double-junction Ag/AgCl references with potassium-free outer bridge electrolyte; no exposed KCl-leaking junction in the measurement cell.

Fluid controlsExact mp6 liquid pump driver, TT Electronics OPB350L062/OCB350L062Z for 1/16 in OD clear tubing, selected fittings, gasket, and closed waste.

Analog completenessSix guarded electrometer channels, bridge excitation/ratiometric acquisition for both MEAS 1620 sensors, ADC input filters, clean analog rails, and leakage-test fixtures.

Independent truthTraceable KCl stock, Class A dilution tools, matrix-matched standards, pH/conductivity checks, and an independent potassium ISE meter or analyzer access.

Chemistry safetyFume-hood access, solvent-compatible PPE and waste, controlled conditioning/storage, and an active/blank membrane recipe made from separate components.

Hydraulic truthThe modeled 0.61 mL channel is acceptable for commissioning, not a fast micro-sample product. At 2 mL/min, three cell volumes alone require about 55 seconds; pump, sensor, and tubing push the valid-output gate beyond one minute. A later cassette should target ≤100 µL active volume if sub-30-second hydraulic refresh is required.

Cost truthThe detailed BOM rows add to $3,285-$7,255, not the older $2.8K-$4.2K headline. Keep a $5K-$10K authorization ceiling only if the integrated carrier, references, driver, controls, shipping, and duplicate wet path are not already covered by placed orders.

Rendered first article

Inspect the bedside CKM architecture.

The 3D preview in the hero and this full viewer show the same first-article assembly: bedrail reader, disposable flow-cell, triplicate K-channel stack, reference/blank controls, KCl standards, pump/flow path, and waste capture.

Open full-screen viewer

Open the full interactive build tourOne full-screen WebGL scene runs more smoothly on mobile and keeps the animated fluid path visible.

Assembly validation

This is the first article to assemble, not just a concept image.

Use this checklist to validate that the ordered parts and fabricated flow-cell match the prototype before testing. Anything that fails here should block chemistry claims until the physical stack is corrected.

01 / CASSETTE

Wet path body

Transparent PMMA/COC/COP body, 120 x 44 x 14 mm nominal envelope, 86 mm long x 3.0 mm ID-equivalent channel, 0.61 mL main-channel volume, and ports for sample, standard, waste, and flush. This is a commissioning geometry pending measured fit and dead-volume review.

02 / CHEMISTRY

Electrode bay

A measured multi-lane carrier must provide K1/K2/K3 and a matched blank in one wetted bay. Existing DRP-110-U75 substrates each provide one 4 mm working electrode; the integrated option is DRP-8X110-U20/CAC8X or a custom six-lane substrate.

03 / READER

Electronics envelope

Reusable reader envelope 170 x 112 x 42 mm with six same-family guarded electrometer channels, ADS124S08 5 x 5 mm ADC, TMP117 local temperature, digital flow sensor, pump/driver, ESP32-S3 data capture, and isolated bench power.

Mechanical and electrical checks before wet testing

Check	Acceptance condition	Why it matters
Channel alignment	Fluid path crosses K1/K2/K3, reference, and blank sensing zone without dead pockets or sharp trapped-bubble corners.	Prevents stagnant fluid and bubbles from looking like K+ movement.
Electrode seating	The selected multi-lane carrier is fixed, sealed, and repeatably positioned over the channel; K1/K2/K3 and matched blank remain wetted together.	Prevents a single-cell development strip from being mistaken for the integrated sensor bay.
Reference isolation	Two independent double-junction Ag/AgCl references use potassium-free outer bridge electrolyte and do not share a failure path.	Prevents reference drift and KCl leakage from becoming false potassium movement.
Guarded analog path	All six high-impedance nodes use one qualified amplifier family, driven guards, clean flux process, short shielded leads, and routing away from pump/display noise before ADS124S08 digitization.	Potentiometric K+ signals are high impedance and can be ruined by leakage.
Flow instrumentation	SLF3S-0600F, mp6 liquid pump plus its high-voltage driver, OPB350 bubble sensor, manual stopcocks, standards, and waste route without kinks or bypass paths.	Validates that every K+ estimate corresponds to fresh, bubble-free known fluid.
Capture readiness	Before first wet run, record raw K1/K2/K3 mV, blank, reference delta, temperature, flow, pump state, sample label, and withheld-output reason.	Makes the first article auditable and reproducible from run one.

Workflow

The buyer is not buying a sensor. They are buying faster potassium control.

Hospital CKM should sit inside existing care pathways: blood gas confirmation, CRRT/dialysis prescription, insulin protocols, diuretic titration, telemetry, and medication review.

Clinical jobs

DKA / insulinWatch rapid intracellular K+ shifts between scheduled BMPs and blood gases.

CRRT / dialysisTrack potassium exchange and dialysate/replacement-fluid prescription response.

Cardiac surgeryDetect drift during post-bypass recovery, diuresis, insulin infusion, and arrhythmia risk windows.

ICU protocolProvide continuous surveillance that triggers a confirmatory sample or protocol escalation sooner.

Hard UI rules

GreenValidated trend, QC pass, recent comparator, stable references, in-range temperature.

AmberTrend usable but confirmation recommended: drift rising, calibration aging, rapid K+ movement.

RedUrgent clinical review: confirmed dangerous trajectory or discordance with comparator.

GrayWithhold output: clot/bubble/flow artifact, reference failure, calibration failure, hemolysis suspicion.

Validation gates

Hospital CKM only ships if it earns trust against existing analyzers.

The hospital bar is higher than the wearable bar. The product must prove analytical performance, hemocompatibility or safe sample isolation, workflow fit, alarm discipline, and clinical usefulness.

AnalyticalK+ slope, bias, precision, drift, hysteresis, temperature response, protein/fouling response, matrix effect, and calibration recovery across 2.0-9.0 mmol/L.

ComparatorDense paired samples against blood gas, i-STAT/POC, central lab, and dialysis/CRRT machine chemistry where applicable.

FluidicNo leaks, no unsafe pressure drop, bubble/clot detection, no hemolysis above accepted thresholds, and no blood-contacting reusable part.

SoftwareFail-closed state machine, audit record, calibration age, comparator age, alarm thresholds, and no autonomous dosing recommendation.

Human factorsNurse installation, sterile boundary, line-change workflow, alarm comprehension, and clinician response must pass simulation before patient feasibility.

First-article acceptance gates

Gate	Pass condition	Unlocks
KCl standards	2.0, 3.0, 4.0, 5.5, 7.0, and 9.0 mmol/L steps produce a stable log-linear response after at least three measured cell volumes.	Matrix-matched dialysate/CRRT testing
Triplicate agreement	K1/K2/K3 agree within the preset mV/K+ window while the blank channel stays quiet.	Confidence score and majority-vote QC
Reference stability	Dual-reference delta stays inside drift limits and no upward K+ creep appears from bridge-electrolyte contamination.	Longer duration runs and calibration aging
Flow artifact rejection	Bubble, stale-flow, and pressure anomalies suppress output instead of creating false potassium movement.	Enclosed reader and valve automation work
Dialysate matrix	Performance holds with matrix-matched calibration in CRRT/dialysate-like sodium, bicarbonate, calcium, magnesium, urea, temperature, and flow conditions.	Artificial plasma and drawn-line blood micro-sample planning

Expected data

The first useful dataset is raw mV truth, not a polished medical number.

Production readiness starts by capturing everything needed to explain every potassium estimate, every suppressed estimate, and every calibration correction.

RAW SIGNAL

Electrode voltages

K1/K2/K3 open-circuit potential versus reference, blank membrane potential, reference A/B delta, ADC noise, settling time, and channel-to-channel agreement.

CONTEXT

Fluidic state

Flow rate, pump state, valve state, upstream/downstream pressure, bubble flag, calibration standard age, sample refresh estimate, and thermal state.

OUTPUT

Validated K+ trend

Temperature-corrected potassium estimate, confidence/QC state, rate of change, last comparator sample, reason for any withheld output, and alarm/confirmation trigger.

Bench expectationValinomycin ISE literature supports Nernstian or near-Nernstian slopes around 55-62 mV/decade, second-scale response in clean solutions, and low drift in controlled setups. The first article should prove repeatable step tracking with production-style QC records.

Hospital barExisting point-of-care electrolyte analyzers already validate the core potassium ISE principle. Hospital CKM must validate continuous flow-cell behavior against blood gas/i-STAT/central lab and must report "confirm now" rather than act as a dosing device.

POC electrolyte performance K+ potentiometric drift example Screen-printed K+ ISE research

Hospital economics

The wedge is high-acuity beds where missed potassium drift is expensive.

This is not a consumer subscription. It is a per-patient disposable plus bedside reader platform for units already paying for blood gas cartridges, dialysis consumables, ICU telemetry, and adverse-event prevention.

ReaderReusable bench module: guarded analog front end, pump control, display, and raw data export. Network integration and battery backup are later requirements.

CassetteNon-clinical flow-cell first; single-patient sterile packaging is a later design-control and manufacturing program.

Use casesDKA, CRRT, dialysis transition, post-cardiac surgery, transplant induction, refeeding, high-risk arrhythmia beds.

ValueEarlier confirmation, fewer hours out of range, fewer unnecessary samples, better protocol adherence.

Build item	Why this part	Planning cost	Scale target
Metrohm DropSens development substrates	Reuse DRP-110-U75 single-cell strips for coating and response screening. Select DRP-8X110-U20 + CAC8X or a custom six-lane carrier before integrated cassette machining.	$285-$650Integrated carrier/connector may require a separate RFQ.	$2-$6 custom printed lane
Valinomycin membrane component set	Cocktail 99373 is a liquid ionophore carrier, not by itself the controlled PVC solid-state membrane. Make active and matched blank formulations from separately controlled ionophore, lipophilic salt, plasticizer, binder, and casting solvent.	$250-$700	$0.50-$2 per cassette chemistry
Solid-contact layer supplies	Use the documented PEDOT:PSS only if its incoming film, conductivity, adhesion, and drift checks pass; otherwise replace it or qualify an MWCNT route.	$120-$450	$0.50-$2 per lane
Two double-junction Ag/AgCl references	Use a potassium-free outer bridge such as 10% KNO3 to stop KCl reference electrolyte from contaminating the low-volume potassium sample.	$250-$650	$1-$4 integrated reference reservoir
Matched non-ionophore blank controls	Use the same binder, plasticizer, salt, thickness, cure, and conditioning as the active membrane, omitting only valinomycin.	$40-$120	<$0.50 chemistry share
Six guarded electrometer channels	Reuse the two documented ADA4530-1ARZ parts for bring-up, then standardize all six K1/K2/K3/blank/reference nodes on the qualified family with driven guards.	$150-$350Guarded layout, cleanliness, and shielding matter more than chip price.	$12-$25 electronics share
TI ADS124S08 24-bit 12-input ADC	Digitizes six buffered electrode/reference potentials plus analog pressure and bubble diagnostics. TMP117 and SLF3S remain digital buses, not ADC channels.	$75-$250Prefer EVM/dev path for first article, then custom PCB.	$5-$12 ADC share
TI TMP117 temperature sensor	High-accuracy local temperature for Nernst compensation and QC gating; 2 x 2 mm package.	$10-$40	$2-$5
Sensirion SLF3S-0600F liquid flow sensor	Orderable microflow validation for sample refresh and valid trend windows; 48 x 15.5 x 8.9 mm.	$115-$205	$20-$45 integrated/OEM
Bartels mp6-liq micropump plus exact driver	Compact controlled flow for the non-clinical standard/sample loop; 30 x 15 x 3.8 mm, 32 µL internal volume, up to roughly 6 mL/min water. The pump requires a high-voltage Bartels driver.	$160-$320	$15-$35 OEM or alternate pump
TE MEAS 1620 pressure sensors	Two -50 to 300 mmHg bridge sensors require stable excitation and ratiometric/PGA acquisition; they are not plug-and-play digital sensors.	$90-$280	$2-$8 disposable share
TT Electronics OPB350L062 bubble sensor	Non-contact optical fluid/bubble detection around 1/16 in OD clear tubing; use OCB350L062Z if the calibrated interface board is preferred.	$30-$120	$0.50-$3 optical share
Manual 3-way stopcocks now; Lee LHD valve bank later	Manual routing reduces first-article complexity. Add automated Lee valves only after chemistry and flow gates pass.	$60-$220Hold automated valves until the gate.	$25-$80 manifolded valves
Two transparent flow-cell bodies, fittings, tubing, KCl standards, waste capture	Physical bench loop: 1/4-28 fittings, PMMA/COC/COP flow body, KCl reservoirs, waste path, and a duplicate wet path.	$1,200-$2,000	$8-$25 molded cassette before sterile pack
Display/MCU/power/enclosure/data-capture harness	Reuse the documented ESP32-S3 and TMP117 breakout where practical; production will require medical isolation and a sealed enclosure.	$450-$900	$55-$120 reusable reader COGS share

Line-item arithmeticThe detailed numeric rows sum to $3,285-$7,255 before shipping, tax, and any integrated sensor-carrier RFQ. A $5K-$10K ceiling is prudent only when placed orders and documented inventory do not already cover those gaps.

At scaleDisposable cassette target COGS: roughly $18-$45 before sterile packaging and QA at low production maturity, with a long-term target around $12-$25. Reusable reader target COGS: roughly $180-$350 early, $90-$180 mature, excluding clinical service and support.

Reconcile before buyingVerify every placed order and packing slip first. The documented TMP117, ESP32-S3, two ADA4530-1 parts, LiPo cells, and conditional PEDOT:PSS are reusable; do not duplicate them. Prioritize only the carrier, potassium-safe references, missing pump driver, bubble sensor, six-channel completion, exact fittings, and independent validation supplies.

Build to shipment

Make the shortest path to a useful hospital pilot.

The right sequence is not "build everything." It is retire the hardest unknowns in the order that creates a credible hospital feasibility package.

Bench CKM flow-cellBuild a transparent cassette with K1/K2/K3, reference, blank, temperature, pressure, bubble, and standard-loop ports. Test in KCl standards, dialysate, artificial plasma, and protein/fouling matrices.

CRRT/dialysate feasibilityProve continuous K+ tracking in dialysate/effluent-like matrices before blood-contacting designs. Show response to prescription-like potassium changes.

Micro-sample blood loopMove to drawn-line sample loops with waste capture. Compare against blood gas/i-STAT/central lab and quantify hemolysis, clotting, drift, and calibration recovery.

Simulation pilotRun ICU nurse and physician simulations: setup, alarm interpretation, comparator prompts, protocol response, and EHR-style event review.

Observational hospital studyNo care-driving claims. Collect paired comparator data, alarm records, setup time, false alarms, and clinician response to trend information.

Boundaries

What Hospital CKM does and does not claim.

DOES CLAIM

Research architecture

A physically plausible hospital potassium trend-monitor architecture with direct access to non-clinical test fluid, redundant K+ sensing, external calibration first, comparator-first validation, and fail-closed workflow.

DOES NOT CLAIM

Clinical readiness

No diagnosis, no autonomous dosing, no replacement of blood gas/lab potassium, no sterile patient use, no mortality reduction claim until analytical and clinical studies prove it.