Cuffless Blood Pressure Wearables: Apple Hypertension Notifications, Samsung, Hilo and What Is Cleared

Short answer: three different things are being sold as blood pressure on the wrist, and only a cuff reading is a clinical measurement.

• A notification: Apple Watch analyzes 30 days of optical pulse data and tells some users they may have hypertension. It gives no number. It is FDA-cleared [2].
• A calibrated estimate: Hilo and Samsung report systolic and diastolic values from the optical pulse signal, calibrated against an upper-arm cuff. Hilo’s is FDA-cleared; Samsung’s US feature is a wellness feature without clearance [4][6].
• A wellness signal: WHOOP and Oura show blood pressure estimates or patterns during sleep, positioned outside device regulation after the FDA’s January 2026 guidance [8][9][10].

Almost half of US adults have hypertension, four in ten of them do not know it, and only one in five has it controlled [1]. That gap is why every large wearable company wants a blood pressure feature. This post covers what each one actually does, how accurate the evidence says it is, and what a health plan, care provider or remote monitoring program can responsibly do with the output.

Why is blood pressure the next wearable battleground?

Because the condition is common, undetected and expensive, and because a cuff is a chore.

CDC data for August 2021 to August 2023 put hypertension at 47.7% of US adults, rising to 71.6% at 60 and over. Among those with hypertension, 59.2% were aware of it, 51.2% were taking medication, and 20.7% had it controlled below 130/80 mm Hg [1]. Awareness, treatment and control did not change significantly from the pre-pandemic survey [1].

A cuff reading is accurate and cheap, but it is a deliberate act at a moment in time. A watch is worn all day and all night. The promise of cuffless measurement is the missing context: nighttime blood pressure, the response to a bad week, the patient who never takes a home reading. The problem is that accuracy standards for that kind of measurement are still being written [11][12].

What does Apple’s hypertension notification do?

Apple’s Hypertension Notification Feature was cleared by the FDA on 11 September 2025 [2]. It analyzes photoplethysmography (PPG) data the watch collects opportunistically during the day, scores each qualifying signal for hypertension risk, and looks across 30-day periods. If the pattern suggests hypertension, it notifies the user. It is for adults 22 and over without a prior diagnosis, not during pregnancy, and it is explicitly not for monitoring treatment or blood pressure surveillance [2].

It never shows a blood pressure number. That is the most misunderstood fact about it.

The validation. Apple enrolled 2,229 adults without a hypertension diagnosis, who wore Apple Watch for 30 days and took readings with an FDA-cleared home cuff. Hypertension was defined as an average of 130/80 mm Hg or higher. 1,863 provided enough data for the primary analysis [2].

• Sensitivity 41.2% (95% CI 37.2% to 45.3%). Most people with hypertension were not notified.
• Specificity 92.3% (95% CI 90.6% to 93.7%).
• Stage 2 hypertension: 53.7% sensitivity. Normotensive specificity: 95.3%.
• Positive predictive value 70.9% at the study’s 31.4% prevalence.
• Older (60 and over) and higher-BMI participants had higher sensitivity and lower specificity; sex, race and skin tone showed no clinically meaningful difference after adjustment [2].

Out of every 1,000 people like the study population, the feature correctly notifies about 129 and stays silent for about 185 who have hypertension. That is the core critique in a 2026 Hypertension commentary: most people with undiagnosed hypertension who wear the watch will not be alerted, and silence can read as reassurance [3]. The authors also note that the FDA cleared it through a 510(k) using a machine-learning ECG product for a heart condition as the predicate, rather than through a blood pressure validation protocol [2][3].

The fair counterpoint is scale. A feature running on tens of millions of watches that finds a third to a half of undiagnosed cases, and is right about 70% of the time when it speaks, will send a large number of people to a cuff who would never have taken a reading. It is a screening prompt. It is not a measurement.

What do Hilo and Samsung measure?

These two report actual systolic and diastolic values, calibrated to a cuff.

Hilo (formerly Aktiia) received FDA 510(k) clearance for over-the-counter use in July 2025, the first cuffless blood pressure monitor cleared for OTC use in the US [4]. The band estimates blood pressure from the optical pulse signal using pulse wave analysis, and is calibrated with an upper-arm cuff that ships with it [4][5]. It launched in the US on 21 July 2026 as Hilo Core at $89.99 plus a $149.99 annual subscription, capturing about 30 to 35 readings a day including overnight, with roughly monthly calibration [5].

Samsung began a phased US rollout of blood pressure monitoring on compatible Galaxy Watches on 31 March 2026 [6]. Users must calibrate with an upper-arm cuff, sold separately, every 28 days. Samsung states that the feature is not intended to diagnose or prevent disease, and it is not FDA-cleared in the US; it launched as a wellness feature consistent with the FDA’s January 2026 guidance [6][7]. Samsung has said passive monitoring of blood pressure trends will follow later this year [6].

Both rely on the same idea. The optical pulse wave changes shape with arterial stiffness and pressure, and a model maps that shape to blood pressure. Calibration anchors the model to one person’s cuff readings, and drift between calibrations is the main accuracy question. The European Society of Hypertension’s validation protocol includes a dedicated recalibration test for exactly that reason [11].

What about WHOOP and Oura?

WHOOP launched Blood Pressure Insights, which estimates systolic and diastolic ranges during sleep from heart rate and blood flow patterns, calibrated with cuff readings. On 14 July 2025 the FDA issued a warning letter calling it a medical device marketed without clearance [8]. WHOOP disputed the letter publicly. On 17 June 2026 the FDA closed it after WHOOP modified the feature, including the dial boundaries that could be read as clinical classification, citing its January 2026 guidance [9].

Oura Ring 5, shipping since June 2026, tracks blood pressure patterns during sleep as part of Health Radar and lets users log cuff readings in the app [10]. It is not presented as an FDA-cleared measurement.

The regulatory shift between those two dates is covered in detail in is your health app a medical device? The short version: the FDA’s revised General Wellness guidance of 6 January 2026 describes a wrist-worn product that outputs blood pressure from PPG as potentially a low-risk wellness product, provided its values are validated [13]. The sensor did not change. The claims around it did.

What do hypertension societies say?

They are consistent, and more cautious than the marketing.

• The European Society of Hypertension published validation recommendations for intermittent cuffless devices in 2023, because existing protocols did not fit them. It defines six tests: static accuracy, device position, a treatment test for falling pressure, an awake and asleep test, an exercise test for rising pressure, and a recalibration test for stability over time. Which apply depends on how the device calibrates and measures [11].
• The American Heart Association’s December 2025 scientific statement says cuffless devices have not yet been adequately vetted for accuracy and efficacy, that standards only recently became available and do not cover typical use, and that current devices may be inappropriate for clinical use, even though several are FDA-cleared and in clinical use [12].
• A 2025 review led by George Stergiou, first author of the ESH recommendations, concludes there is no convincing evidence that any cuffless technology has the accuracy required for clinical use. It adds that tools like Apple’s notification can help detect undiagnosed hypertension but need cuff confirmation, and are not intended for people already being treated [14].

Two things follow. First, FDA clearance and clinical acceptance are different bars; a cleared device can still fall short of what a hypertension society would use to make a treatment decision. Second, the most defensible use of cuffless data today is finding people and showing patterns, not diagnosing or titrating medication.

What can a health plan or RPM program do with it?

On remote patient monitoring billing: CMS requires the device used for remote physiologic monitoring to meet the definition of a medical device in section 201(h) of the Federal Food, Drug and Cosmetic Act. CMS has said the CPT codebook does not require FDA clearance, but that clearance may be appropriate [15]. A feature its own manufacturer says is not intended for medical use is a weak foundation for an RPM claim. From 1 January 2026 Medicare also pays for shorter monitoring periods, 2 to 15 days of data in 30 days under code 99445, alongside a 10 to 19 minute management code, 99470 [16]. That makes episodic monitoring with a connected, cleared cuff easier to justify. The broader shift is covered in remote patient monitoring going consumer.

How should a product store and show these readings?

The mistake to avoid is a single blood pressure series that mixes sources.

• Keep provenance on every value. Source device, method (cuff, calibrated cuffless, notification), and for cuffless values the date of the last calibration. A reading 27 days after calibration is not the same quality as one taken the day after.
• Never average across methods. A cuff reading and a wrist estimate are different measurements. Chart them separately, or show cuff readings as anchors over a cuffless trend.
• Model notifications as events, not values. Apple exposes a hypertension event type in HealthKit from iOS 26.2 [17]. It says something happened; it does not carry a number.
• Use the platform types that exist. HealthKit has systolic and diastolic blood pressure types, and Health Connect has a blood pressure record [18][19]. Which readings a vendor writes there, and whether it marks them as estimates, varies by vendor and changes over time, so check the source metadata rather than assuming.
• Word the prompt, not the diagnosis. “Your readings have been higher than usual; consider a cuff reading” is a wellness prompt. “You may have hypertension” is a disease claim that Apple needed FDA clearance to make.

The short version

Apple’s FDA-cleared feature notifies, it does not measure, and in its own study it flagged 41% of people with hypertension while being right about 71% of the time when it did [2]. Hilo’s FDA-cleared band and Samsung’s US wellness feature produce calibrated estimates that depend on a monthly cuff reading [4][6]. WHOOP and Oura sit on the wellness side of the FDA’s January 2026 line [9][10][13]. Hypertension societies say none of it is yet accurate enough for clinical decisions [11][12][14]. Use it to find people and show patterns, confirm with a cuff, and never let the three kinds of number share one chart.