Samay

The sensing method has no non AI interpretation path. The device emits low frequency sound into the chest and reads returning tissue resonance, and machine learning converts that acoustic signal into lung structure and function measures including air trapping, lung volume, capacity, and flow rates. Unlike spirometry, which produces a directly interpretable measurement, acoustic resonance is meaningless without the model, so the algorithm is the instrument.

Output is presented for clinician assessment, with results surfaced through a companion app and a research portal for pulmonologists and primary care physicians to review, and the stated intent is to let clinicians intervene early rather than to automate a decision. Because the platform is investigational rather than cleared, the practical oversight position today is that all output sits in a research or monitoring context and cannot be relied on diagnostically. No published detail describes alerting thresholds or escalation logic for detected deterioration.

The physical method is described in unusual detail, covering active acoustic resonance, the patented approach of projecting sound and reading returning signal, and the pairing of lung physiology with IoT sensors, digital signal processing, and machine learning. The company has also presented method and results in a respiratory medicine conference abstract. The model itself is not described, and no algorithm documentation or validation methodology beyond headline accuracy figures was located.

Nothing identifies any party in the chain: no model or model family, no hosting arrangement and no sub processor list was located in two passes, and no retention, encryption or consent terms were found. One question specific to this device outranks all of those, and no other wearable in this index raises it. The sensing mechanism is acoustic.

The device projects sound into the chest and reads the returning resonance using speakers and microphones, worn taped to the chest and described as operating around the clock while the person goes about ordinary life, with data moving to a companion application and onward to backend infrastructure and care teams. Company material also lists lung sounds among the metrics collected.

An always on body worn microphone in a domestic setting captures more than the lung: it sits inches from the wearer's larynx and in the same room as everyone else in the household, none of whom has consented to anything. Whether that matters turns entirely on where the acoustic signal is processed and what is retained.

A device that extracts resonance features on board and discards the waveform raises very little; one that streams or stores raw audio for later analysis raises a great deal, including for third parties in the room. Nothing published states which it is, and it is answerable in a sentence. Ask whether raw audio ever leaves the device or is retained anywhere, and put the answer in the protocol and the consent document.

Stronger evidence than most companies at this stage, and appropriately caveated. A 110 patient study conducted with a Florida pulmonary institute across COPD, asthma, and healthy controls reported detection of air trapping at 83 percent accuracy against hospital pulmonary function tests and COPD diagnosis at 90 percent accuracy, with statistically significant associations in medication efficacy pre and post bronchodilation.

Results were presented as an abstract in the American Journal of Respiratory and Critical Care Medicine, and the company received NIH small business program support. Buyers should weigh three limits: cohort size is modest, the abstract is not a full peer reviewed paper, and the study used a prototype device.

Converted from Not Rated. No retention, encryption or consent terms were located, and a second search surfaces a stewardship question specific to this device that no other wearable in this index raises.

The sensing mechanism is acoustic. The device projects low frequency sound into the chest and reads the returning resonance using speakers and microphones, worn taped to the chest and described as operating around the clock while the person goes about ordinary life, with data moving to a companion phone or tablet application and onward to backend infrastructure and care teams. Company material also lists lung sounds among the metrics collected.

An always on body worn microphone in a domestic setting captures more than the lung. It sits inches from the wearer's larynx and in the same room as everyone else in the household, none of whom has consented to anything. Whether that matters turns entirely on where the acoustic signal is processed and what is retained: a device that extracts resonance features on board and discards the waveform raises very little, while one that streams or stores raw audio for later analysis raises a great deal, including for third parties.

Nothing published states which it is. That single question, whether raw audio ever leaves the device or is retained anywhere, should be the first one a study sponsor, a health system or a participant asks, and it is answerable in a sentence.

The rest of the ordinary gaps remain. No statement of retention periods, encryption in transit or at rest, who at the company can access participant data, whether data from a study is used to train or refine models beyond that study, or what happens to a participant's data on withdrawal. For a device intended for continuous home use and for endpoint capture in decentralised trials, all of that belongs in the protocol and in the consent document.

Converted from Not Rated. A scoping determination, and it closes for the company's current engagements while leaving a clearly marked boundary.

The device is investigational and not cleared, and the engagements described are research studies and endpoint capture for pharmaceutical trials. In that setting the instruments that govern are the study protocol, institutional review board approval and the participant's informed consent, with the sponsor or the site holding the regulatory relationship to the participant. A vendor business associate agreement is not the operative document, and its absence is a correct consequence of the stage rather than an omission.

The boundary is worth stating precisely because the company's stated ambition crosses it. The positioning is to replace effort dependent spirometry and to enable continuous home monitoring for chronic obstructive pulmonary disease and small airway disease. Continuous home monitoring ordered by a clinician for a patient's care is not research; it is treatment, and at that point the vendor is receiving identifiable health information on behalf of a provider and the framework changes completely. Nothing published addresses that configuration.

A health system evaluating this for monitoring rather than for a study should therefore treat this row as answering a different question from the one it is asking. Establish whether the company will execute a business associate agreement, which entity signs, and what changes in the data handling when a participant becomes a patient.

Converted from Not Rated. No SOC 2, ISO 27001 or equivalent attestation was located and there is no trust centre, which is consistent with the company's stage and is recorded without implying more than that.

What the stage does not excuse is the attack surface, which is broader than a software product's and is worth naming while the architecture is still being decided. This is a wireless body worn sensor communicating with a consumer phone or tablet, which relays to backend infrastructure that shares metrics with care teams. That chain has three distinct exposures: the wireless link between device and phone, a companion application running on a device the company does not control, and a backend holding continuous physiological data for a clinically vulnerable population. A medical device connected to a personal phone also inherits whatever else is on that phone.

For a company at this stage the useful question is not whether an attestation exists but what is being built toward. Ask whether the device link is authenticated and encrypted, whether the application stores participant data locally on the phone and for how long, whether the backend is segregated per study, and whether an attestation is on the roadmap ahead of any clinical deployment. A pharmaceutical sponsor placing this in a decentralised trial will require most of that contractually regardless.

Search caution for anyone re running this. The company appears in at least one consortium listing under a different corporate name, with the same founder, device and description. Where a company has operated under more than one name, an attestation or certificate may sit under either, so search both before concluding that nothing exists.

No FDA clearance is evidenced and the device is described in prototype and development terms across sources, including a peer reviewed abstract stating the company is developing the wearable to facilitate COPD diagnosis. This is the central constraint on the product: a device measuring lung function and diagnosing COPD is squarely within device regulation, so absent clearance its use is confined to research, clinical trial endpoint capture, and investigational monitoring. NIH small business program funding is research support rather than any form of regulatory standing. Buyers should treat every published accuracy figure as investigational.

Converted from Not Rated. The prior note asked whether signal quality and model accuracy vary with body habitus, chest wall thickness, sex and comorbidity, since acoustic resonance depends on tissue properties. Published work now answers part of that question, and the answer raises a larger one that a buyer must put to the company directly.

The company's published diagnostic classifier takes demographic information as model input alongside the device signal, specifically age, sex and race. Race is therefore an input variable in a model that outputs a respiratory diagnosis.

That is not automatically wrong, and it should not be read as an accusation. It is, however, the single most contested methodological question in this exact clinical area. Pulmonary function interpretation used race specific reference equations for decades, and the field has since moved decisively away from them, on the finding that race adjustment systematically normalised worse measured lung function in some populations and thereby delayed diagnosis and affected decisions about disability, transplant candidacy and occupational compensation. A device whose stated purpose is to replace effort dependent spirometry, and whose model takes race as an input, has reintroduced the variable the field has spent years removing from the test it aims to replace. Whatever the justification, it needs to be stated, and none was located.

The accompanying disclosure gap compounds it. The published evaluation reports aggregate discrimination, sensitivity and specificity on a validation subset of sixty six subjects, and reports no performance breakdown by race, sex or body habitus, despite race being an input. Where a demographic variable is used to generate a prediction, subgroup performance is not an optional refinement; it is the minimum evidence that the variable is doing what the model assumes.

Two further method points to raise. The reported split of the dataset into training and validation is described as random, and where each participant contributed several recorded manoeuvres a random split at recording level can place the same person on both sides. And no external validation on an independent cohort was located.

What would move this row: the justification for including race, performance reported by subgroup, and a validation cohort the model has not seen.

The physical method is described in unusual detail and the inferential layer on top of it is not. Active acoustic resonance is explained concretely, covering the patented approach of projecting low frequency sound into the chest and reading the returning signal, paired with connected sensors, digital signal processing and machine learning, and the company has presented method and results at a respiratory medicine conference, which is a venue where the approach met an audience able to question it.

A clinician can therefore understand what physical quantity is being measured, which is more than most wearables in this index permit. Held at C because the step from resonance signal to clinical output is undocumented. No model class, no validation methodology beyond headline accuracy figures, no per condition operating characteristics and no warranty, indemnity or remediation commitment were located, so a buyer knows how the sound is produced and read and nothing about how a pulmonary conclusion is drawn from it.

The intended use raises the stakes on that gap: a device positioned for continuous home monitoring and for endpoint capture in decentralised trials produces numbers that either trigger a clinical response or become trial data, and both uses need a stated error characteristic. Ask for per condition sensitivity and specificity against a reference standard, and how the device behaves on poor signal quality.

Converted from Not Rated. A scoping determination, and the architecture is appropriate to what the company is currently doing rather than deficient against a standard that does not yet apply.

The stack is a chest worn device, a paired mobile application and a research portal. That is a study oriented architecture: it moves signal from a participant to a research team, which is the interoperability the company presently needs. No electronic health record integration is claimed and none would serve the current use, since an investigational device generating research measurements does not write to a patient's chart and should not.

Grading the absence of a clinical integration that the product is not designed to have, at a stage where having one would be inappropriate, would misdescribe the vendor. The domain equivalent question is whether study data reaches the people who need it, and a research portal answers that.

The axis becomes live the moment the positioning the company states is realised. Continuous home monitoring for chronic respiratory disease, used clinically, means measurements a clinician acts on, which means those measurements belong in the record with provenance, timing and device identification attached, and it raises the alerting question that every remote monitoring vendor in this index has to answer: who receives an abnormal reading, how fast, and what happens if nobody does. Nothing published addresses any of that.

So read this grade as covering the research architecture only. A buyer evaluating clinical monitoring should ask for the integration roadmap, the record write path, and the alert routing design, and should expect that none of it exists yet.

Converted from Not Rated. The deployment surface is described and the hosting position is not.

What is described: a chest worn device paired wirelessly to a participant's own phone or tablet, a companion application that performs signal analysis, a research portal, and backend infrastructure through which respiratory metrics reach care teams. That is a four part chain, and the first two parts sit on hardware the company does not own, one of which belongs to the participant.

What is not described: where the backend runs, in which region, whether study data is segregated per sponsor or per site, retention, subprocessors, or whether the analysis happens on the phone, in the cloud, or across both. The last of those is the one that determines everything else, because it decides what actually leaves the participant's home.

The question is becoming more consequential rather than less. The company's stated direction includes high frequency endpoint capture for pharmaceutical trials and replacing routine site visits in decentralised studies. A decentralised trial moves the data collection point into the participant's home and the data itself into a sponsor's evidence package, so residency, segregation between sponsors, and the chain of custody from device to case report form all become regulated concerns rather than engineering preferences.

Ask where the backend sits, whether processing is on device or server side, how one sponsor's study data is isolated from another's, and what the data path is from the sensor to a sponsor's dataset.

No published pricing. Third party commentary has framed the device against the cost of CT scanners and formal pulmonary function testing, and the company positions three commercial contexts, primary care early detection, home monitoring, and pharmaceutical trial endpoint capture, but no rates or structures are disclosed.

Narrow by design and clearly stated: respiratory, specifically COPD and small airway disease, with studied populations spanning COPD, asthma, and healthy controls. Three intended settings are named, primary care early detection, continuous home monitoring, and high frequency endpoint capture in pharmaceutical clinical trials. The trial endpoint use case is the one least constrained by the absent FDA clearance and is therefore the nearest term commercial path.