Nanowear

Learned models are the engine of the platform's analytics, and the cleared garment has value without them. SimpleSense is an undergarment of cloth nanosensors that records two ECG leads, respiration rate by thoracic impedance, heart sounds and activity, which is the cleared indication for the platform (K212160) and is directly readable by a clinician.

On top of that the company sells AI enabled analytics in reports for physicians, and its cleared blood pressure product (SimpleSense-BP, K232053, December 2023) is an ensemble of models that estimates systolic and diastolic pressure from features of the garment's signals and the patient's demographics, a measurement that exists only as an inference. The company describes itself as building deep learning based diagnostics from more than 120 million data points per patient per day. Remove the models and the garment still records cleared physiological data, so the model is the engine of a core module rather than the whole product.

The boundary is published in regulatory language and the operational rules are not. The company's security page publishes the platform's indication for use: to record, display and store two ECG leads, respiration rate, heart sounds and activity, under the direction of a licensed medical professional.

The blood pressure filing (K232053) adds that the estimates are retrospective, produce no alarms, are not for active patient monitoring, require recalibration with a cuff every 21 days, and are rejected when movement is detected. Output reaches physicians as reports.

What is missing is operation: who reviews the reports in home use, what triggers escalation, how quickly, and what the AI enabled real time analytics the product page describes do between reports, which sits awkwardly beside a filing that rules out active monitoring.

The sensing and one model are described specifically. The garment uses cloth based nanosensors in a gender neutral, size adjustable undergarment that the company says captures electrophysiological, hemodynamic, acoustic and activity signals time synchronously.

The SimpleSense-BP filing (K232053) describes an ensemble of models estimating systolic and diastolic pressure from features extracted from garment data plus age, gender, height and weight, states that the algorithm is fixed rather than adaptive, and requires cuff recalibration every 21 days; company material has described it as four decision trees.

The deep learning behind the AI enabled real time analytics in physician reports is not described, with no model, operating point or version, and the company's description of a platform that gets smarter with each patient is not reconciled with the fixed model in the filing. The company holds 13 awarded and 12 pending patents across sensors, manufacturing, platform, ingestion and algorithms.

The chain is structurally short at the model layer and partly named around it. The blood pressure model is Nanowear's own, cleared as a fixed algorithm (K232053) that cannot be changed without regulatory change control, and the company's patents cover its sensors, manufacturing, platform, ingestion pipeline and algorithms, so no third party model provider appears in published material.

The security page names Amazon Web Services as host, and the company has named partners whose data or products meet the platform: Dexcom for continuous glucose readings, a blood collection patch developer and a surgical device distributor. No subprocessor register or contract manufacturer is published, and nothing states the retention and training terms that govern data exchanged with partners or the consent basis for any model learning from deployed data.

One product has published validation with method, and the headline claim is much broader. The SimpleSense-BP summary (K232053) reports two studies under IEEE 1708: induced blood pressure change in 147 usable subjects across hypertension categories (mean difference 0.09 mmHg, SD 4.08 systolic) and a 21 day calibration study of 91 subjects with no evidence of drift.

The product page claims the garment captures more than 120 million data points a day and replaces the blood pressure line, digital stethoscope, multi channel ECG, hemodynamic monitor, spirometer and actigraphy monitors; the evidence published answers blood pressure estimation, not that claim.

The company has named a remote diagnostic research alliance with a New Jersey hospital system and a Brooklyn medical center and a distribution partnership for post surgical recovery, with no deployment volume or outcome figures, and a 60 percent workflow reduction quoted by an executive carries no method.

Security controls are described and the AI questions are not answered. The company's security page says data is hosted on Amazon Web Services and encrypted in motion and at rest (HTTPS, AES-256), with role based access controls, 24/7 monitoring, redundant systems and encrypted backups replicated offsite, and states a commitment to the privacy of protected health information under federal and state law.

Nothing states a retention schedule, whether patient recordings are used to develop models, or how data is de-identified. Company material has described the platform as getting smarter with each patient, while the blood pressure filing (K232053) describes a fixed model; which models learn from deployed data, and under what consent and change control, is unstated. The record is dense: the company says the garment captures more than 120 million data points per patient per day. No safety engineering is published.

A HIPAA commitment is stated and the document behind it is not shown to reach the service. The security page says Nanowear is committed to maintaining HIPAA compliance and, as a participant in patient health care, to the privacy of protected health information under applicable federal and state law, and it links the company's privacy policy as its notice. No business associate status, agreement, execution requirement or subcontractor terms are published.

The platform is cleared for home, facility and research use, which produce different contracting pictures, and the Dexcom data partnership moves patient data across a company boundary on terms not described.

Controls are asserted with nothing independent behind them yet. The security page lists AWS hosting, AES-256 encryption in motion and at rest, role based access controls, 24/7 monitoring, redundancy and encrypted offsite backups, and says Nanowear is working toward SOC 2 Type II certification with Vanta. No completed attestation, penetration testing on a stated cadence, vulnerability disclosure route or trust portal is published.

The surface is a connected device worn against the body in patients' homes, sending a dense continuous record and exchanging data with a partner's glucose system, and the blood pressure product is cleared software with cybersecurity expectations of its own; device authentication, firmware practice, garment decommissioning between patients and the partner data channel are unaddressed.

The clearances are real and the marketing claims a status they do not carry. FDA's database lists four 510(k) clearances for Nanowear, Inc.: SimplECG (K161431, November 2016) and SimpleSENSE (K201669, November 2020) and the SimpleSENSE Platform (K212160, September 2021), all product code DXH (ECG transmitters and receivers, 21 CFR 870.2920), and SimpleSense-BP (K232053, December 2023, product code DXN, non invasive blood pressure, 870.1130).

The company's own security page publishes the indication: to record, display and store two ECG leads, respiration rate by thoracic impedance, heart sounds and activity including posture, under a licensed professional's direction; the blood pressure filing adds retrospective estimates with no alarms, not for active monitoring.

The product page calls SimpleSense an "FDA cleared remote diagnostic platform" that replaces the blood pressure line, digital stethoscope, multi channel ECG, hemodynamic monitor, spirometer and actigraphy monitors, monitors lung volume, and generates AI enabled real time analytics; no Nanowear clearance covers spirometry, lung volume, hemodynamic monitoring or real time diagnostic analytics. The security page also still says three clearances.

The blood pressure model's evaluation discloses its groups and counts without results by group. The SimpleSense-BP 510(k) summary (K232053, 8 December 2023) describes an induced change study of 149 enrolled subjects (147 usable; 68 female, 81 male; 17 Asian, 45 Black, 20 Hispanic, 67 White) with at least 21 subjects in each hypertension category from normal to stage 2, and a 21 day calibration study of 91 subjects, about 74 percent hypertensive, and the company describes training and validation in independent geographies with populations mirroring the national census.

That is a subgroup design named with counts; no accuracy is reported by sex, race, body habitus or blood pressure category, so where the model underperforms cannot be seen. The cardiopulmonary analytics carry no comparable statement. Sensing depends on cloth contact, so signal quality plausibly varies with body habitus, breast tissue, chest hair, skin condition and fit. No fairness framework, drift monitoring or outside audit is published.

Nanowear's own filing publishes an error figure for its learned blood pressure model with method and case counts. The SimpleSense-BP 510(k) summary (K232053, cleared 8 December 2023, Nanowear, Inc.) describes an ensemble model estimating systolic and diastolic pressure from garment signal features and demographics, fixed rather than adaptive, validated under IEEE 1708: in an induced change study of 147 usable subjects against auscultatory reference, mean difference 0.09 mmHg (SD 4.08) systolic and 0.35 mmHg (SD 3.32) diastolic, within the ISO 81060-2 criteria; in a 21 day calibration study of 91 subjects, weekly systolic mean differences of -2.94 to -0.88 mmHg (SD 4.82 to 5.13).

These are test results for adults at rest with a stationary torso, recalibrated with a cuff every 21 days, for retrospective estimates that raise no alarms; they do not cover the cardiopulmonary analytics, and they are not performance in use. No warranty, indemnity or remediation commitment is published, and where analyses combine the company's data with a partner's glucose readings, nothing says how responsibility divides.

Device level data exchange is named and record system integration is absent. The company has described a licensing and data partnership that ingests Dexcom continuous glucose readings time synchronously with its own cardiovascular data, and an initiative to integrate blood collection patches, and it holds a patent on its ingestion pipeline. The product page says output reaches physicians as easy to read reports. No electronic health record is named, no HL7 or FHIR interface is described, no marketplace listing appears, and nothing states whether recordings or reports reach the patient's chart.

The host is named and its location is not. The security page says the platform runs on Amazon Web Services with redundant systems, no single point of failure, and encrypted backups replicated offsite. No region, residency commitment, tenancy model or recovery objective is published.

The platform is cleared for use at home, in a healthcare facility and in medical research organizations, which involve different networks and data paths, and nothing says whether deployment differs or whether a facility can keep data on premises. For home use the garment depends on connectivity the company does not control, and nothing states whether it buffers locally or what happens to a recording interrupted mid session.

The company publishes nothing that sizes the cost: no pricing page, unit of charge, range, tier, implementation fee or reimbursement route. The product is a garment worn against the skin, so whether it is bought, leased, replaced per patient, laundered and reused, or supplied per diagnostic episode decides the economics, and sizing, fitting and reprocessing are operational costs that no material addresses. The site has a cart page whose contents are not described on the product page. The cleared indications sit where remote monitoring and diagnostic billing codes plausibly apply, and no billing route is published.

Coverage is named with validation behind part of it. The cleared settings are home, healthcare facilities and medical research organizations under a licensed professional's direction, and the blood pressure product is validated in adults 22 and older at rest across hypertension categories (K232053, 147 and 91 subjects). Cardiology and cardiopulmonary assessment are the established domain, with post surgical recovery through a distribution partner.

The product page's wider claims, replacing spirometry, hemodynamic monitoring and other instruments, and the company's named target conditions (COPD, sleep apnea, worsening heart failure) have no published validation, and the adjacent markets it names (sports, industrial safety, government) sit outside healthcare. Fit varies with body habitus, mastectomy, ostomy, wound sites and mobility, which nothing addresses.