Biobeat

The sensor is the product, not the model. Apply this index's standard test: remove the AI early warning score and what remains is a cleared seven parameter continuous wearable monitor with a cloud platform, which is a complete and saleable product in its own right.

That is the same test that led to the iFax rejection, and Biobeat is NOT rejected on it for two reasons stated here for the record: it is healthcare specific rather than a horizontal product with a health tier, and the early warning layer does address this category's actual job rather than sitting decoratively on top. Graded C with the mechanism stated plainly. What Biobeat genuinely contributes to this category is measurement, and that contribution is real.

An automated early warning score generating real time deterioration alerts, with no published threshold, no operating characteristics, no abstention or uncertainty behaviour, and no adjunctive use statement retrieved. Users can also set their own threshold alerts on vital signs, which places some control with the institution, but nothing describes who governs the AI derived score or what a clinician is expected to do when the device based score and the hospital's existing early warning process disagree.

In a category whose defining failure is alerts being ignored, a vendor adding a further automated alert stream to a ward carries a specific burden to show what its alert volume looks like, and nothing published addresses it.

A sharp split between two halves of the same product.

The sensor side is well described: a proprietary patented reflective photoplethysmography sensor, the derivation of seven distinct physiological parameters from that single optical signal, and clear statements of which parameters carry clearance and when each was granted.

The AI side is the vaguest claim in this category. The early warning score is described only as incorporating advanced AI based algorithms, with no model class, no inputs specified beyond the device's own measurements, no scoring basis, no thresholds, no validation, and no statement of whether it maps to any published early warning framework such as NEWS or MEWS.

This axis grades the model, and there is effectively nothing to grade. Sensor transparency is recorded as the counterweight but does not lift the grade.

Nothing identifies any party in the chain: no model or model family, no hosting provider, no sub processor list, and no privacy policy, data processing statement or retention period was located in two passes. Three features of this record make that gap wider than the usual absence.

The data is dense: these monitors capture on the order of thirteen to fifteen physiological parameters continuously, including pressure derived optically, oxygen saturation, respiratory rate, cardiac output and a single lead electrocardiogram.

A continuous multi parameter stream is not a set of readings, it is a physiological record rich enough to support inference well beyond the purpose it was collected for and distinctive enough that individual identifiability becomes a real question rather than a theoretical one. The geography is unstated: the company is headquartered in one jurisdiction and sells into the United States, Europe and Latin America, so processing location and transfer mechanism are live and unaddressed.

And the data moves onward, flowing into at least one third party care at home platform under a partnership, which introduces a further party into the path without naming what it receives or retains. A consumer channel would place the same stream under a different regime again. Ask where data is processed and retained, the transfer mechanism, what the partner receives, and whether patient data trains the scoring models.

The regulatory record implies measurement validation, since cuffless blood pressure derived from photoplethysmography required accuracy demonstration against reference standards to clear, and successive clearances for additional parameters imply the same each time. That is real, but it is not what this axis measures well.

No peer reviewed outcome study was retrieved, and no named hospital deployment with published results. Critically, no validation of the early warning score itself was located: no sensitivity, no positive predictive value, no alert burden, no lead time, and no comparison against the conventional early warning scores it implicitly competes with.

For a product whose deterioration claim is the reason it belongs in this category, the deterioration claim is the least evidenced part of it.

Converted from Not Rated after a second search. No privacy policy, data processing statement, retention period or training use disclosure was located. The prior note's framing holds and the second pass adds two things.

The first is what the data actually is. These monitors capture on the order of thirteen to fifteen physiological parameters continuously, including blood pressure derived from optical measurement, oxygen saturation, respiratory rate, cardiac output and a single lead electrocardiogram. A continuous multi parameter stream is not a set of readings; it is a physiological record dense enough to support inference well beyond the clinical purpose it was collected for, and dense enough that individual distinctiveness becomes a real question rather than a theoretical one.

The second is where it goes. The company is headquartered in Israel and sells into the United States, Europe and Latin America, so processing location and transfer mechanism are live and unstated. Data also flows onward into at least one third party care at home platform under a partnership, which introduces a further controller or processor into the path. And the consumer channel puts the same data under a different regime entirely.

The early warning capability adds a training question. A deterioration score improved from the streams it observes implies that patient data contributes to model development, and nothing states whether it does, under what basis, or whether a provider can decline.

Ask where data is processed and retained, for how long, what the transfer mechanism is, what the partner receives, and whether patient data trains the scoring models.

Converted from Not Rated after a second search. The company states consistently across its own materials that its cloud patient management platform is compliant with both the United States health privacy rule and European data protection law. That is a published claim and it is more than silence, but it names two regimes rather than describing a posture: no business associate availability statement, contracting entity, permitted use terms or subprocessor list was located.

The finding that matters here is structural and the second search surfaced it. The same wearable monitors are sold through two quite different channels. In the clinical channel a hospital or care at home programme deploys them and views data through the platform, which puts the company in a business associate relationship with that provider. In the consumer channel the company sells a chest monitor directly to individuals from its own website, with self set up requiring no clinician and results delivered through an app to the person themselves. That second path does not involve a covered entity at all, so the health privacy rule does not reach it and the data sits under consumer protection law, state consumer health data statutes and, for European users, data protection law.

One device, two legal regimes, and nothing published distinguishes them. A patient cannot tell which applies to them, and a hospital cannot assume its contract covers a device its patient bought independently.

Ask which entity signs, what covers the consumer channel, and how data captured in one channel is separated from the other.

Converted from Not Rated after a second search. No SOC 2, HITRUST, ISO 27001, trust centre or report request path was located. The prior note's distinction holds and is worth restating: the company's medical device quality audit certification is a quality management programme, not an information security attestation, the same separation this index applies to device quality standards, conformity marking and clearance throughout.

A further point now applies and it runs against the vendor rather than for it. This index credits recently cleared devices for the cybersecurity documentation that submissions must now carry, because a security plan, a software bill of materials and a post market update process became submission requirements in 2023. The clearances located for this company all predate that: cuffless blood pressure and the core vital signs earlier, respiratory rate and temperature in early 2022, stroke volume and cardiac output in January 2023. None therefore carries that review, and clearance here says nothing about security.

What sits behind the gap is a continuously transmitting estate. Body worn devices stream vital signs from patients in hospitals, care at home programmes and consumer settings across several regions, into a cloud platform, with onward integration into at least one third party care at home platform. The device firmware, the transmission path, the platform and the partner integration are four surfaces and none is described.

Ask what independent security examination exists, how devices authenticate to the platform, how firmware is updated, and what the partner integration exposes.

A genuinely strong and progressively built regulatory record. Biobeat states its devices were the first ever cleared by the FDA for cuffless blood pressure monitoring derived from photoplethysmography alone, an August 2019 clearance covering continuous blood pressure, blood oxygen saturation and pulse rate. Clearance was extended in March 2022 to respiratory rate and body temperature, and again in October 2024 to stroke volume and cardiac output, giving seven cleared parameters from a single sensor. Add CE marking and MDSAP certification for international market access. Deriving haemodynamic parameters such as stroke volume and cardiac output from an optical wearable is a substantive clearance rather than a formality.

The scope precision matters and is recorded here rather than left implicit. These clearances cover measurement of physiological parameters. Nothing retrieved indicates the AI based early warning score is itself cleared, so a buyer reading FDA cleared alongside AI based deterioration alerts should not conflate the two. Ask which components the clearances cover, and under what regulatory basis the early warning score operates.

The sharpest instance of the photoplethysmography pigmentation problem anywhere in this index, and it is structural rather than incidental.

Pulse oximetry is well documented to overestimate arterial oxygen saturation in patients with darker skin pigmentation, a measurement bias the FDA has itself examined, and the mechanism is optical: reflective photoplethysmography measures light interaction with tissue through skin. The same question raised against Etiometry applied to one input among several. Here every parameter is derived from photoplethysmography. Blood pressure, oxygen saturation, pulse rate, respiratory rate, temperature, stroke volume and cardiac output all trace back to one optical sensor, so a pigmentation dependent measurement error does not affect a single reading. It propagates through all of them and into any early warning score computed on top.

Nothing published addresses this. There is no subgroup accuracy data by skin pigmentation, no calibration analysis, and no statement of the populations in which the clearance validation was performed. Publishing accuracy by skin pigmentation across the cleared parameters would be the single highest value disclosure this vendor could make, and until it exists the question should be put directly in any evaluation.

Two passes located no validation of the early warning score, no thresholds, no scoring basis, no statement of whether it maps to any published early warning framework, and no warranty, indemnity or remediation commitment. That is the vaguest artificial intelligence claim encountered in this category, described only as incorporating advanced algorithms, and it matters because the score is the thing clinicians act on: a deterioration score is what triggers an escalation, a rapid response call or a decision to leave a patient where they are.

One distinction on this record deserves naming because it is a certification transfer problem occurring inside a single product rather than between companies. The sensor side is well described and specific parameters carry clearance, with the dates stated. Those clearances cover the measurement of physiological parameters.

They do not cover the inference drawn from them, and an early warning score derived from cleared measurements is a different regulatory and evidentiary object from the measurements themselves. A buyer reading a list of clearances alongside a score should not assume the second inherits the assurance of the first.

The two failure directions are also asymmetric in the usual way: a score that fires too often is silenced by staff, and a score that misses a deteriorating patient produces an event nobody attributes to the monitor. Ask what the score is derived from, what thresholds trigger what action, and for validation against a recognised framework.

Integration into the hospital EMR is claimed and described as seamless, but no EHR vendor is named anywhere retrieved, no FHIR or SMART on FHIR capability is described, no marketplace listing or partner certification was located, and no integration architecture beyond cloud transmission is published.

The interoperability position is also structurally different from the rest of this lane and worth understanding: because the device generates its own data rather than consuming existing feeds, Biobeat needs far less integration to function at all, but correspondingly it produces a parallel data stream that must be reconciled with whatever the hospital already charts, and nothing published describes how measurements flow into the record or how conflicts with cuff based readings are handled.

The deployment property is genuinely differentiated and it is the practical argument for this product. Every other vendor in this category requires integration with infrastructure the hospital already owns, whether an EHR feed, a central monitoring system or a device estate, which is precisely what excludes facilities without that infrastructure. Biobeat brings its own sensor, so a ward can be continuously monitored without a monitoring estate existing first, using a disposable patch or wrist device and a cloud platform the company describes as fast and easy to install. That materially lowers the barrier for general wards, community hospitals and hospital at home programmes.

Held at B rather than A because no named hospital deployment was retrieved anywhere, no implementation timeline or scale reference exists, and no data residency commitment is published, which is a live question for cloud hosted continuous telemetry crossing borders. Named international distribution partners cover the Dominican Republic, Argentina and Chile.

No pricing published at any level: no device cost, no per patient or per patch price, no platform subscription, no band and no implementation fee. The consumable model makes this a more consequential gap than for a software only vendor, since a disposable chest patch implies recurring per patient cost that scales with monitored volume, and nothing published indicates what that is or how the wrist device is priced by comparison.

One partial disclosure is noted without being credited: the company refers to a clear reimbursement pathway for ambulatory blood pressure monitoring, but no specific code or payment rate was retrieved, so this falls short of the reimbursement transparency that earned Eko and HeartSciences credit elsewhere in this index.

The widest setting span in this category, and deliberately so. Marketed coverage runs across general wards, surgical wards, post surgical recovery, step down units, medical surgical units, intensive care and emergency departments, then continues outside the hospital into ambulatory blood pressure monitoring, nursing homes and hospital at home programmes.

That continuity is the product thesis rather than a list, since the same wearable follows a patient across acuity levels and out of the building, which is exactly what a discharge or step down decision needs and what no other vendor assessed here offers. Parameter coverage is also broad at seven cleared measurements from one sensor.

Held at B because the cleared population is adults with no paediatric or neonatal indication retrieved, and because breadth of marketed setting is not the same as evidenced performance in each of them, none of which is published.