How to verify AI wearable blood pressure monitor accuracy

The uncomfortable part of buying an AI wearable blood pressure monitor is that the most official-sounding phrase on the box may answer the wrong question. The American Heart Association put it plainly in its December 2025 scientific statement: “FDA clearance for a blood pressure device does not require formal accuracy testing under a standardized protocol. Therefore, regulatory clearance does not guarantee measurement accuracy.” [1]

That does not make every smartwatch, ring, band, or cuffless sensor useless. It does mean a buyer has to separate three claims that often get blended together: the product is authorized for some BP-related function, the product has passed a recognized accuracy protocol, and the product uses AI to estimate trends or trigger alerts. Those are different claims. Before relying on readings, the public record has to show which one is actually true.

The quick verification path

For a pre-purchase check, do not start with the brand’s words. Start with the device name, model, app feature name, and country where you intend to use it. Then work through the public evidence.

A vague phrase such as “clinically tested,” “doctor recommended,” “AI-powered,” or even “FDA-cleared” is not the endpoint. It is the prompt to look up the record.

Step one: check the FDA record, not just the FDA language

The FDA’s September 16, 2025 safety communication warned consumers not to use unauthorized devices for measuring blood pressure, including software features on smartwatches and smart rings. The agency told consumers to check the FDA 510(k) database and identified product code DXN for blood pressure devices. [2]

That lookup is worth doing slowly because “FDA-cleared” can be true and still incomplete. A 510(k) record can show that a device or feature has a U.S. clearance pathway and can identify the applicant, device name, decision date, and product code. It is the right place to confirm whether the product is actually authorized in the U.S. for the BP-related claim being advertised.

It is not the same as an accuracy report. The AHA statement is the guardrail here: FDA clearance does not, by itself, mean the device completed formal standardized accuracy testing. [1]

When you search, use the most specific terms you can find: the marketed product name, the manufacturer name, the feature name, and product code DXN. If a company sells a smartwatch with several health features, do not assume one clearance covers every sensor, every country, or every app screen. The record you want is the one that corresponds to the blood pressure measuring or monitoring function you are being asked to trust.

• If the device appears under product code DXN, treat that as evidence of U.S. authorization status for the listed device or feature, then continue to accuracy validation.
• If the company only says “FDA registered,” “listed,” or “compliant,” do not read that as clearance for blood pressure measurement.
• If the record is for a different model, a different accessory, or a different function, do not transfer that status to the wearable you are buying.
• If the product is cleared or approved in another country, treat that as jurisdiction-specific evidence, not proof of U.S. FDA clearance.

This jurisdiction point matters with wearable BP features. Samsung’s Galaxy Watch blood pressure feature has been discussed in connection with non-U.S. regulatory pathways, including CE marking and Korean MFDS approval, while Consumer Reports describes the U.S. situation differently and notes a 28-day recalibration requirement for Samsung’s wearable BP approach. [3] A buyer in the U.S. should not treat a feature available abroad as if it has the same U.S. medical-device status.

Step two: look for the exact device on validatebp.org

After the FDA check, use validatebp.org as the buyer-friendly shortcut. Consumer Reports points readers there when discussing wearable blood pressure devices, and Cedars-Sinai also directs consumers toward it when explaining what smartwatch hypertension detection can and cannot do. [3][4]

The useful habit is to search for the exact device, not the brand family. A brand can sell one validated upper-arm cuff, one unvalidated wrist cuff, and one smartwatch feature that is not meant to produce diagnostic blood pressure readings. Validation does not automatically travel between them.

Absence from validatebp.org should not be exaggerated into proof that a device is fraudulent. It does mean the buyer should ask for another public, named validation source. A private “clinical study available on request” is weaker than a device appearing in a public list or being tied to a recognized protocol with enough detail to identify what was tested.

Step three: translate the standards instead of memorizing the acronyms

The standard is where a marketing phrase either gets sharper or falls apart. “Clinically validated” should lead to a named protocol, not a mood. For conventional noninvasive BP validation, AAMI/ESH/ISO criteria are commonly summarized around mean error within ±5 mmHg and standard deviation within 8 mmHg; ISO 81060-2 and ISO 81060-3:2022 are part of the standards landscape that separates “tested somehow” from “tested under a recognized protocol.” [5]

Cuffless devices need special care because they are not simply smaller cuffs. IEEE 1708-2025 is the active IEEE standard for wearable cuffless blood pressure measuring devices; it was board-approved on November 4, 2025, published on April 3, 2026, and supersedes IEEE 1708-2014. Its purpose is objective performance evaluation for cuffless BP devices. [6]

A 2026 review compiling cuffless BP standards also lists ISO 81060-3:2022 and an ESH 2023 cuffless protocol with six evaluation scenarios, which is a useful reminder that cuffless validation is not one generic lab comparison. The conditions under which a wearable is tested matter. [5]

For a product page, the practical question is simple: does the company name the protocol and connect it to the exact wearable? If the answer is only “AI model trained on millions of signals,” that may describe software development, but it does not answer measurement accuracy against a recognized BP standard.

Why this extra work is not nitpicking

The market gives buyers plenty of reasons to check. A 2025 Clinical Hypertension paper cites that 75–80% of home blood pressure devices and 100% of the studied cuffless devices were not validated for accuracy. [7] That does not prove any one device you are considering is inaccurate. It does show why a buyer should not treat a wearable BP claim as verified until the specific device clears the public-evidence checks.

There is also a difference between measuring blood pressure and warning about possible hypertension. Apple Watch Hypertension Notification is a useful contrast case because it is an FDA-cleared notification feature, not a cuffless blood pressure monitor that gives systolic and diastolic readings. Consumer-facing explanations note that the feature requires about 30 days of wear before a notification can appear. [4]

The much-circulated “92% accurate” framing around Apple’s notification feature also needs the denominator attached. In the research brief’s source set, that figure refers to positive predictive accuracy when an alert fires, while 58.8% of undiagnosed hypertensive people receive no alert. Those two facts describe different things: one is about the likelihood an alert corresponds to hypertension, the other is about missed undiagnosed hypertension among people who do not get alerted. They should not be merged into “the watch is 92% accurate at detecting hypertension.” [4]

That boundary matters for someone already monitoring hypertension at home. A passive alert can be useful as a prompt to take a proper measurement or talk to a clinician. It is not the same thing as a validated blood pressure reading, and it should not be treated as permission to stop using a validated cuff.

How to read common claims before buying

The strongest product evidence usually has boring details: exact model names, public database records, named standards, study conditions, and clear limits. The weaker evidence leans on polished phrases that sound official but avoid saying what was measured.

A practical status judgment

Use a dated status label for the device you are considering. Regulatory status and validation evidence can change, so the check should be repeated before purchase and again before relying on a new software feature.

Verification date for this guide: August 25, 2026. This article is not medical advice, does not diagnose hypertension, and does not authorize medication changes based on wearable readings or notifications. If a wearable raises concern, use a validated blood pressure monitor as directed and involve a qualified clinician.

References

1. Cuffless blood pressure technologies in wearable devices show promise to transform care, American Heart Association Newsroom, December 2025
2. Do Not Use Unauthorized Devices for Measuring Blood Pressure: FDA Safety Communication, U.S. Food and Drug Administration, September 16, 2025
3. Measuring Blood Pressure With a Wearable Device, Consumer Reports
4. Can a Smartwatch Really Detect Hypertension, Cedars-Sinai
5. Cuffless Blood Pressure Monitoring Based on Wearable Sensors: A Review, Nano-Micro Letters, January 2026
6. IEEE 1708-2025, IEEE Standards Association, April 3, 2026
7. Clinical Hypertension 2025 Galaxy Watch 6 blood pressure study, Clinical Hypertension, 2025