BioIntelliSense

Genuine analytics performing the work that makes the product viable, on top of a data capture layer that delivers real value without any inference at all.

The base proposition is measurement rather than intelligence. Automating vital sign capture removes a manual task, eliminates the estimation bias that makes charted respiratory rates cluster implausibly at round numbers, and stops waking patients for observations. A hospital would gain from that with no models involved, and much of the published economic argument rests on it.

What the models add is the reason the data is usable. Thousands of measurements per patient per day across a whole floor is an unmanageable volume for human attention, and the exception management layer is what converts it into fewer than one notification per patient per day. Personalised trending thresholds mean the system learns what is normal for an individual rather than applying population cut offs, and deterioration pattern detection is what turns a data stream into an earlier intervention. Without that layer the product would drown the unit it was bought to help.

The intelligence is therefore load bearing without being the whole structure, which is the definition this index applies at this grade.

Graded B rather than higher because no accuracy or operating point is published for the deterioration detection, so the analytic claim cannot be examined, and rather than lower because the alert volume figure demonstrates the filtering is doing something real.

The two numbers that determine whether continuous monitoring works in practice are both published, and they are published together.

Alert fatigue is the recognised failure mode of this entire category. Systems that generate more signals than a unit can action teach staff to dismiss them, and the dismissal is invisible until something is missed. This vendor publishes fewer than one notification per patient per day and clinician engagement with those notifications at 99.71 percent. Either figure alone would be incomplete: a low alert rate could mean the system is insensitive, and a high engagement rate could reflect a small sample. Together they describe a system whose signals are rare enough to be credible and are actually being acted upon.

The mechanism behind them is described rather than asserted. Exception management with personalised trending thresholds per patient means notifications fire on deviation from an individual's own baseline rather than on population cut offs, which is why the volume stays low without simply raising the bar.

The operating model is human throughout. One clinician monitors many patients through a configurable dashboard and decides what to act on, and the flagship deployment runs a centralised command centre with a multidisciplinary team, which is a named and staffed escalation structure rather than an assumed one.

What remains unpublished is the miss side. No sensitivity for deterioration detection is stated, so the proportion of deterioration events that produce no notification is unknown, and a low alert rate is only reassuring alongside that figure.

Ask for deterioration detection sensitivity, and the staffing ratio behind the command centre model.

The measurement layer is specified and quantified, while the analytic layer that creates the value is described only by name.

What is disclosed is concrete. Sensor placement is stated as the upper left chest, the measured parameters are enumerated as heart rate at rest, respiratory rate at rest, skin temperature, body position and activity, capture volume is quantified as thousands of multiparameter measurements per patient per day, and the architecture is named component by component across the wearable, the wireless gateway, the cloud analytics and the dashboard. Operational performance is quantified where it matters most, with alert volume and clinician engagement both published.

Independent publication supplies figures the company did not. An academic validation reported sensitivity and specificity for activity classification and 78.6 percent accuracy for body position, including a weak result of 0.52 sensitivity against accelerometer reference for moderate to vigorous activity. A buyer can therefore find genuine, unflattering performance data on some functions.

The deterioration analytics remain unspecified. No sensitivity, specificity, lead time, threshold logic or validation is published for the pattern detection that constitutes the clinical claim, no model card exists, and how personalised baselines are established is not described.

That asymmetry is the finding. Vital sign accuracy is evidenced; the intelligence built on top of it is not.

Ask for deterioration detection performance and lead time, and how personalised thresholds are derived and adapted.

A dedicated pass located no cloud or infrastructure provider, no sub processor register, no device manufacturer, no sensor or silicon supplier and no machine learning component disclosure.

The cloud omission is certain rather than probable, since the analytics layer is explicitly cloud based and named as a product component, so a hosting provider exists and is unnamed. A health system therefore cannot determine whose infrastructure holds continuous physiologic data on its inpatients.

The hardware chain is equally undisclosed and this is a device company. Sensors, wireless gateways and charging stations are physical products with manufacturers, component suppliers and firmware origins, and none is identified. For connected medical devices deployed at the scale of several thousand beds in a single system, provenance is a security and continuity question rather than a curiosity.

One element of the chain is partly visible through corporate history. A previously independent clinical intelligence platform was acquired and folded into the portfolio, which means part of the analytic stack originated outside the company, and nothing describes how it was integrated or whether it retains separate infrastructure.

Training provenance is unaddressed. The deterioration analytics were developed against physiologic data from somewhere, the company holds monitoring data on more than 400,000 patients, and no statement connects or separates the two.

Ask for the hosting provider and sub processor register, the device manufacturer, and the provenance of the data behind the deterioration models.

An outcome study, an independent validation the company did not commission, and a full system deployment reference.

The outcome study is the strongest element. Peer reviewed research in the Journal of Clinical Medicine covered nearly 12,000 hospitalised patients across the medical surgical units of two hospitals over 15 months, reporting reduced length of stay, low alert notification rates and earlier clinical intervention on deterioration. Length of stay is an outcome rather than an accuracy measure, and 12,000 patients over 15 months is real world operation rather than a pilot.

The independent validation matters more than its subject suggests. An academic study evaluated the device's activity and position detection in children against reference measurement and published sensitivity and specificity, and the results were mixed rather than flattering, including sensitivity of 0.52 against an accelerometer for moderate to vigorous activity alongside strong performance for sedentary behaviour and 78.6 percent accuracy on body position. Independent evaluation that reports unflattering numbers is the most credible evidence a vendor can have associated with it, precisely because the vendor did not control the result.

Deployment evidence is at system scale and named. A large health system completed expansion across all 2,653 non intensive care beds at eight hospitals with a centralised command centre, which is a full commitment rather than a unit trial, and more than 400,000 patients had been monitored as of October 2025.

The qualification is study design. The outcome study is retrospective and observational, so secular improvement and concurrent initiatives cannot be excluded.

Ask for the length of stay effect size and comparator, and deterioration detection performance.

A dedicated pass located no encryption statement, no retention schedule, no access control model, no audit logging description, no deletion process and no position on whether patient data contributes to model development.

The dataset this product creates is the reason the silence matters. A continuous multiparameter physiologic record captures when a patient slept, how much they moved, how their position changed through the night and how their vital signs trended hour by hour, at a resolution no chart has ever held. Across more than 400,000 patients that is a substantial and unusually granular corpus.

The training question follows directly and is unaddressed in both directions. Deterioration detection algorithms improve with exactly this kind of labelled longitudinal data, the company holds it at scale, and nothing states whether patient monitoring data has been used for model development, under what consent, or whether a health system can decline.

The device layer adds its own exposure. Body worn sensors communicating to wireless gateways across a hospital network are connected endpoints in a clinical environment, and nothing describes device authentication, pairing security or what happens to a device that leaves the building on a discharged patient.

One pre emptive note: further clinical evidence cannot move this grade. Only a published retention schedule, a data ownership position and a training data statement will.

Ask what the cloud retains and for how long, who owns it, whether it trains models, and how devices authenticate.

Two dedicated passes located no health privacy position of any kind: no compliance statement, no business associate agreement template, no execution requirement and no description of which entity contracts.

The architecture makes the omission substantive rather than formal. Continuous physiologic data travels from a body worn sensor through wireless gateways installed on the hospital network to cloud analytics, so identifiable patient data leaves the institution continuously and by design. This is not a device that processes locally and reports a result; the cloud layer is where the analytics live.

The home deployment complicates it further and is unaddressed. When the same device supports hospital level care at home, data traverses a patient's domestic network rather than a hospital's, and nothing describes how that path is governed or what the patient is told.

The volume is the aggravating factor. Thousands of measurements per patient per day across 400,000 patients constitutes one of the larger continuous physiologic datasets held by any vendor in this index, and no retention period, deletion process or ownership statement accompanies it.

Customers are large health systems with mature procurement, so agreements demonstrably exist across the base and simply are not published, which leaves a prospective buyer unable to begin diligence without a sales conversation.

Ask for the agreement template, what the cloud layer retains and for how long, who owns the resulting dataset, and how home deployments are handled.

Two dedicated passes located no security page, no external attestation, no trust centre, no penetration testing statement, no vulnerability disclosure policy and no documentation offered under agreement. No security credential of any kind was found.

The surface is larger than a software platform's and includes physical devices in patient contact. Body worn sensors, wireless gateways distributed through hospital buildings, charging stations and a cloud analytics layer together form a connected medical device fleet operating inside clinical networks. Device authentication, firmware update practice, pairing security and decommissioning are all first order questions for that architecture and none is addressed.

One scenario is specific to the reusable model and unaddressed. A rechargeable multi patient device cycles between patients through charging stations, which makes the sanitisation and data separation process between admissions both a hygiene and an information security question.

A second follows from the disposable variant, since a single patient device worn home by a discharged patient is a connected sensor that has left the institution entirely.

The buyer profile makes the absence more surprising rather than less. Customers include large academic health systems that run demanding vendor security reviews, so a posture certainly exists in private, and none of it is published where a prospective buyer could begin diligence.

One pre emptive note: further clinical evidence cannot move this grade. Only an external attestation, or documentation available under agreement, will.

Ask whether an external security assessment exists, how devices authenticate and update, and how reused devices are cleared between patients.

Multiple clearances covering both the wearable and the software that interprets it, with the detail of each unpublished.

The October 2024 clearance is the most informative because of what it covered. The rechargeable multi patient wearable and the clinical intelligence dashboard were cleared together as a system, which means the software presenting and filtering the data was assessed alongside the sensor producing it. That is the right regulatory shape for this product, since the exception management layer is what a clinician actually acts on, and a cleared sensor feeding an uncleared dashboard would leave the consequential half unexamined. Earlier clearances cover the single patient device, and an acquired clinical intelligence platform carried its own clearance.

The positioning is also correctly bounded. This is monitoring and trending for general ward patients rather than a diagnostic claim, and the company does not describe it as detecting named conditions.

What holds the grade below the imaging records in this index is disclosure and reach. No clearance numbers, indications for use or classification are published, so a buyer cannot establish precisely what each device is cleared to measure or what claims are permitted. No European or other international marking was located, so this appears to be a single jurisdiction product.

One question follows from the analytics. Whether the deterioration pattern detection is within the cleared indication, or offered as an informational trend outside it, is not stated.

Ask for the clearance list with indications, and whether deterioration detection sits inside the cleared claim.

A dedicated pass located no fairness testing, no subgroup performance, no calibration data, no drift monitoring, no governance framework and no external algorithmic audit.

The bias exposure here is specific, well documented in the clinical literature, and entirely unaddressed. Body worn sensors of this type derive cardiopulmonary measurements optically, and optical measurement through skin is affected by pigmentation. Pulse oximetry has an extensive published record of overestimating oxygenation in patients with darker skin, with consequences serious enough to have prompted regulatory review, and the same physics applies to optical sensing generally. A chest worn device measuring heart rate and respiratory rate optically sits squarely in that literature, and no performance data by skin tone was located.

The consequence in this setting is direct. If measurements are less accurate for some patients, then personalised trending thresholds are built on a noisier baseline for those patients, and deterioration detection degrades for exactly the group whose outcomes are already worse.

A second axis concerns body habitus. Sensor contact, adhesion and signal quality vary with chest size, tissue composition and body hair, and nothing addresses performance across those variations.

The independent paediatric validation demonstrates that stratified evaluation is feasible for this device, since an academic group published subgroup relevant performance the company had not.

One pre emptive note: further deployment scale cannot move this grade. Only performance stratified by skin tone will.

Ask for measurement accuracy by Fitzpatrick skin type and by body habitus, and whether such analysis was submitted for clearance.

Nothing contractual is published, for a system that hospitals substitute for an existing observation regime.

A dedicated pass located no service level agreement, no accuracy warranty, no uptime commitment, no indemnity and no remediation position.

Substitution is what makes this different from an additive monitoring tool. A ward adopting continuous automated capture reduces manual observation rounds, so the new system is not layered on top of the old practice but replaces part of it. When that happens, availability is a patient safety property rather than a service quality one, and no uptime commitment or historical availability figure is published anywhere.

The detection side carries the familiar gap. Published alert volume and engagement figures characterise how the system behaves when it fires, and no sensitivity figure describes how often it should have fired and did not, so a hospital cannot quantify the deterioration events the programme will miss.

The home deployment adds a third dimension. Hospital level care at home depends on a domestic network the vendor does not control, and nothing states where responsibility sits when connectivity fails and a patient is unmonitored at home under the hospital's clinical accountability.

One pre emptive note: further deployments or publications cannot move this grade. Only contractual terms, or published availability and deterioration detection performance, will.

Ask for the availability commitment and historical uptime, what is warranted on detection, and how connectivity failure is handled in home deployments.

The documentation burden is central to the value proposition, and no record system integration is specified anywhere.

The architecture is described in its own terms. Wearables communicate to wireless gateways, gateways feed cloud analytics, and results surface in the vendor's dashboard, with a large health system operating a centralised command centre against that dashboard. Within its own boundaries the system is coherent and its components are named.

What is absent is the connection outward. No electronic health record is named, no interface standard is described, no marketplace or validated integration listing was located, and nothing states whether captured vital signs post to the chart as discrete, timestamped, trendable observations.

That gap sits awkwardly against the stated benefit. Company material identifies the burden of repetitive manual tasks associated with spot check measurement and documentation as the problem being solved, and documentation is only solved if the values reach the record automatically. If a nurse still transcribes readings from a dashboard, the measurement half is automated and the documentation half is not.

A second consequence follows for clinical continuity. Continuous physiologic trends that live only in a vendor dashboard do not reach the discharge summary, the next admission or any retrospective review of whether the programme changed outcomes.

Graded C rather than lower because the internal architecture is genuinely specified and a command centre integration is demonstrated at scale.

Ask which record systems are integrated in production, through what standard, and whether vital signs post as discrete charted observations.

The physical topology is described component by component, and the hosted layer behind it is not described at all.

What is published is more than most. The path runs from a body worn sensor to wireless gateways installed within the facility, then to cloud analytics and a dashboard, with charging stations supporting the reusable device variant. A hospital can therefore understand what equipment it must site, that gateway coverage is a prerequisite across monitored areas, and that wireless network capacity is part of the deployment. That is practical information a facilities and network team can act on.

Everything on the vendor side is absent. No cloud or hosting provider is named, no region is stated, no residency commitment is made, no tenancy or segregation model is described for a platform serving multiple health systems, and no backup or continuity position was located.

Continuity deserves particular attention for this product. When a unit has replaced periodic manual observation with continuous automated capture, an outage does not degrade monitoring gracefully; it removes the observation regime the ward has reorganised around, and nothing describes fallback behaviour or availability history.

The home deployment raises a further unaddressed question, since hospital level care at home moves the gateway onto a domestic network with no institutional support.

Ask which provider hosts the analytics and where, how tenants are segregated, what happens during an outage, and how home gateways are supported.

Cost is absent from every published surface, alongside an unusually well sourced argument about somebody else's cost.

A dedicated pass located no pricing page, no unit of charge, no range, no tiering, no implementation fee position and no minimum commitment. Nothing indicates whether charging follows the device, the monitored patient day, the bed, the unit or the enterprise, which matters more here than usual because two device variants exist with different economics: a disposable single patient sensor consumed per admission, and a rechargeable multi patient device with charging stations that amortises across many patients.

What is published is a comparator argument built with citations rather than assertions. Continuous telemetry is stated to cost hospitals up to 1,400 dollars per patient per day, a retrospective review is cited finding 22.3 percent of patients inappropriately assigned to telemetry at admission with over half monitored beyond guideline duration, and a 2024 multicentre study is cited attributing 33 percent of patient sleep interruptions to routine vital sign checks. Sourcing the inputs rather than asserting a return figure is the right practice and this index credits it.

It does not substitute for a price. A hospital can see that the alternative is expensive and misapplied, and still cannot calculate what this costs or what it displaces, and the capital footprint of gateways and charging infrastructure is likewise unstated.

Ask for the charging unit, how the disposable and rechargeable variants price against each other, the gateway and charging infrastructure cost, and the contract term.

Coverage designed around one clinical population and extended across the places that population moves through.

The target is the general ward patient, and the settings named follow that patient rather than scattering: medical surgical units, specialty care areas, emergency departments, hospital level care delivered at home, and post acute environments. That is a coherent span because the same monitoring gap exists in all of them. Intensive care is deliberately outside scope, since those patients are already continuously monitored, and the flagship deployment is explicitly across non intensive care beds.

Hardware supports the range rather than the marketing claiming it. A disposable single patient device suits home and post acute use where return logistics are impractical, while a rechargeable multi patient device with charging stations suits inpatient units where the same device cycles through admissions. Building two form factors is evidence the settings were engineered for.

The independently validated paediatric use extends the population, though as research rather than a marketed indication.

What is missing is distribution and geography. More than 400,000 patients is a cumulative figure that says nothing about how many sites run this today, what proportion are inpatient against home, or whether any deployment exists outside the United States. No international regulatory marking was located.

Ask for the split between inpatient and home deployments, the current site count, and whether paediatric use is a cleared indication.