Wearable blood pressure research: what PPG shows about trends and vascular aging

Blood pressure is the most requested measurement in wearable health and one of the hardest to get right. Of the 152 wearable studies on our map of wearable PPG research, 25 focus on blood pressure and another 18 on arterial stiffness and vascular aging. Together they are the largest block of work on the map. This review reads that literature with one question in mind: what can a team running a study or building a product honestly learn about blood pressure from a light sensor on the skin?

The research is large and the validation bar is high. Some blood pressure questions suit wearable PPG today and some do not, and this page sorts them.

For a plain introduction to the consumer side of the topic, see this guide to blood pressure watches. This page focuses on the research and on study design.

Why blood pressure is hard to read from light

PPG does not sense pressure. It senses changes in blood volume under the skin, which rise and fall with each heartbeat. Blood pressure has to be inferred from how that pulse behaves. Research has used two broad routes.

Pulse timing. The pressure wave from each heartbeat travels along the arteries. When pressure rises, arteries become stiffer and the wave travels faster, so the time it takes to reach a sensor gets shorter. Pulse transit time is measured between two points on the arterial tree. Pulse arrival time adds the delay between the heart's electrical signal and the pulse reaching the sensor, so it is usually measured with ECG plus PPG. A detailed review of the theory showed why the link between timing and pressure is real but personal: it depends on the size and stiffness of each person's arteries and on smooth muscle tone, which changes during the day 1. You can read more about the timing idea in this explainer on pulse wave velocity.

Pulse shape. A single PPG sensor cannot measure transit time between two points, so single-site methods look at the waveform itself. Features include the rise time, the height and timing of the reflected wave, the dicrotic notch, and the second derivative of the waveform. Machine learning models then map those features to cuff values. A review of single-site PPG for blood pressure, covering 2010 to 2019, described a fast-growing set of feature and model choices with mixed results 4.

The VascAgeNet review grouped methods into three families: a single PPG signal, multiple PPG signals or transit time, and PPG combined with other signals such as ECG for arrival time 2. The family matters, because each one fails in different ways.

What the wearable blood pressure studies tested

The 25 blood pressure studies on the map span lab benches, clinics and a few free-living recordings. A few patterns stand out.

Most studies are short and calibrated. A 2016 review traced continuous measurement from invasive catheters to unobtrusive sensors and described how calibration works in practice 12. In a widely cited evaluation, 33 people were measured with a pulse-transit-time method that added a second PPG feature. Errors grew when people did maneuvers that changed vessel tone and grew again as the time since calibration lengthened 5. That pattern, good agreement right after calibration and drift afterwards, shows up across the field.

Single-site deep learning is promising but hard to judge. PP-Net, a deep learning model, estimated blood pressure and heart rate together from PPG alone 13. Papers like this report results on public datasets, which helps comparison but often mixes recordings from the same person into training and testing.

Larger comparisons tend to be spot checks. One study compared a PPG-based device with a standard cuff in 1,057 people at a screening station 14. That is a large sample, but each person gave a short reading at one moment, which says little about tracking change over hours or weeks.

A few studies went ambulatory. A pulse-arrival-time system was compared with a cuff over 24 hours in 24 people 15. Work like this is closest to what a wearable could add, and it is still small.

An npj Digital Medicine review of PPG for hypertension concluded that the technology was not yet mature for clinical use and called for larger, more diverse validation 3.

Why calibration and validation decide everything

If a method needs a cuff reading to start, and then drifts, the cuff is doing part of the work. A 2025 review in Hypertension found no compelling evidence that pulse wave analysis or arrival time add value beyond the cuff calibration or simple demographics such as age and sex, and noted that many accuracy claims came from weak testing procedures 8.

Standards bodies have responded. IEEE published a standard for wearable cuffless devices in 2014 16. Experts later argued that the validation protocols used for cuff devices are not suitable for cuffless ones, because a single session cannot show drift or the effect of treatment 17. The European Society of Hypertension issued a statement that it does not recommend cuffless devices for the diagnosis or management of hypertension 6, followed by recommendations for how they should be validated 7. Those recommendations describe six tests:

1. A static test against reference readings.
2. A position change test.
3. A test after a change in treatment.
4. An awake versus asleep test.
5. An exercise test.
6. A recalibration test, to see how long accuracy holds.

Few published PPG studies pass all six. For anyone designing a study, that list is a useful checklist. It also explains why research claims about "measuring blood pressure" should be read carefully. The hypertension overview here and this piece on when cuffless measurement may arrive cover the same validation debate for a general audience.

Trends, not readings: the night-time question

There is a different question that suits wearable data better. Instead of asking for a number at a single moment, it asks how pressure-related signals change across a day and night.

Blood pressure normally falls during sleep. Researchers sort people by how much it falls, using terms such as dippers, non-dippers and reverse dippers. The labels date back to a 1988 Lancet letter by O'Brien and colleagues 18. Since then, large cohorts using cuff-based ambulatory monitoring have linked the night-time pattern with outcomes. In the Ohasama study of 1,542 adults followed for about nine years, each 5 percent smaller night-time decline was associated with roughly 20 percent higher cardiovascular mortality 9. The ABC-H meta-analysis of 17,312 people with hypertension found that a reverse dipping pattern was linked with higher rates of cardiovascular events 10. European guidelines give weight to out-of-office measurement, including ambulatory monitoring, which is where these night-time patterns come from 19.

Wearables are worn overnight, and PPG features change with vessel tone. One study used a digital PPG reflection index and found higher vascular tone at night in 31 people with essential hypertension than in 30 controls, rising through the night 20. That does not turn a wearable into an ambulatory monitor. It does suggest a research path: track relative overnight change in PPG timing and shape features within the same person, then test those trends against cuff-based ambulatory monitoring.

This is the frame we use for blood pressure on our own pages. We talk about trend categories and relative change in research, not readings. Sensor Bio does not report blood pressure values.

Vascular aging

Arterial stiffness and vascular aging are the second half of this cluster, with 18 studies on the map. Here the evidence is older and more consistent, mostly because the questions are simpler. Instead of estimating a moving number, researchers ask whether the pulse shape reflects how stiff the arteries are.

The second derivative. Takazawa and colleagues showed in 1998 that the second derivative of the finger PPG waveform changes with age and responds to drugs that change vessel tone 11. Later work in 973 middle-aged men found that age, along with factors such as exercise habits and blood lipids, was independently linked with the same indices 21.

Pulse timing at different sites. In 116 people, age was the strongest factor in pulse timing measured at the ears, fingers and toes 22. Multi-site timing is closer to a true transit-time measure, which is why multi-sensor studies dominate this area.

Comparisons with reference methods. Digital PPG stiffness indices have been tested against applanation tonometry. In 112 people, several PPG indices were repeatable and correlated with augmentation index and pulse wave velocity 23. A waveform index from finger PPG correlated strongly with aortic augmentation index in a study of healthy people and people with type 2 diabetes 24. Another comparison in 83 people found a more modest correlation between a PPG stiffness index and tonometry 25. In 152 healthy volunteers, PPG stiffness and reflection indices were correlated with age, heart rate and blood lipids, and race was an independent predictor of the stiffness index 26.

The VascAgeNet review summarized the field: PPG carries real information about vascular age, but methods are not standardized, most data come from fingers at rest, and wrist PPG needs its own validation 2. For a deeper look at the methods, see this overview of PPG vascular age assessment.

For research teams, vascular aging is often a better fit for a wearable than blood pressure. The question is about slow change over months or years, measured at rest, where wrist motion matters less.

What this means for research teams

If your study touches blood pressure or vascular health, the literature suggests a few design choices.

Ask a trend question. Relative overnight change, response to an intervention, or slow change in pulse shape over months are all better matched to wearable PPG than single readings.

Keep a reference. Pair the wearable with cuff-based ambulatory monitoring or tonometry for at least part of the study. The six ESH tests are a useful template for when to collect it 7.

Keep the raw waveform. Pulse shape features, second derivatives and beat timing come from raw samples, not summary metrics. As an example, Sensor Bio V2 records raw green, red and infrared PPG at 25, 50 or 100 Hz, and partner teams can export those waveforms for their own analysis. V2 is a general wellness device and is not FDA cleared, and it does not report blood pressure.

Log context. Posture, sleep and wake times, activity and temperature all change vessel tone. Without them, a trend can be an artifact.

Check signal quality first. Shape features are sensitive to weak or noisy signals. See PPG signal quality for how quality checks change results.

How these signals are measured: the science of PPG. For every research area, return to the wearable PPG research map.

For research teams

Put the raw signal in your study or product

Planning a study on overnight trends or vascular aging? We can help you plan raw PPG capture and export.

Talk to Sensor Bio

Limits and open questions

• Most blood pressure studies are small, short and calibrated, so they test agreement soon after calibration rather than tracking over time.
• Wrist PPG is noisier than finger PPG, and most vascular aging work used fingers.
• Few studies report results across skin tones, ages and body types in enough detail.
• Single-site methods cannot measure true transit time, and their added value beyond calibration is still debated 8.
• The link between overnight PPG feature trends and cuff-based dipping patterns has not been established in large studies.