Can you measure blood pressure in a smart watch?
April 2025 , Sam Moreland
There has been talk about blood pressure being performed in many different places; a toilet seat, behind the ear or your smartphone camera. But the primary cause of interest recently has been from Apple and Asus who are claiming that they can/will be able to measure blood pressure from a watch, without a cuff. The reality is they most likely can’t with enough clinical accuracy, but it could do something else.
Now first and foremost, getting a continuous and cuff-less measure of blood pressure would be hugely valuable to society as a whole. 24/48 hour ambulatory BP measurements can be very painful for older adults and the completion rate for a full measurement can be quite low. On top of that, getting a true diurnal (difference between day and night) understanding for BP is very important and night time BP measurements with current technologies can often wake people up.
Companies can produce smart watches very cheaply and healthcare is very lucrative, so they want to combine the two together. And with the advent of AI, there is a lot of hype going around with not much understanding of the physiology of how blood pressure is measured and why it's one of the hardest vitals to innovate with. I hope to go through in the article and explain the ins and outs of the most difficult vital.
A few things to keep in mind.
As we go through the article there are a few things you should think from a clinical point of view.
Your BP is normal at 120 / 80 mmHg (120 systolic, 80 diastolic) and depending on your circumstances, you could be put on hypertensive medication when your systolic is 140 mmHg. The most accurate BP cuffs in the world give 8 mmHg std error. You can think of this as meaning that if you have a systolic of 135 mmHg, you have a 26.6 % chance for it to read over 140 mmHg. And likewise, if your systolic is 145 mmHg, you have a 26.6 % for it to read under 140 mmHg. The error rate is quite high even with the best equipment.
Your BP can vary very quickly, just standing up can increase your systolic BP by 10-20 mmHg and it can take 5-10 minutes (or longer if you're older or have heart issues) to come down. Walking and running will have drastically larger changes than this. Thats why when assessing your BP you really should sit still for 5-10 mins after any activity before you take a reading.
There is a massive difference between estimation and measurement. Estimation is when you make a judgement about non-direct variables to make a judgement about a variable. For example, I don’t have windows in my house, if a person comes in and is wearing a raincoat, I could estimate that it has been raining (and do it fairly accurately if the coat was wet). Measurement is where you directly observe the different levels of a variable, for example I use a thermometer to measure the temperature of my food.
Companies will use words to make it seem that they can measure things they cannot. This is because regulators are very strict on what you can say when marketing a medical device, however this is not widely known in the public. Because of this, companies will use words like detect or estimate, some will even say measure when they are not cleared to (this is illegal), in order to make you think it is a qualified medical product.
Why do I need a cuff?
We are trying to measure something non-invasively that is perpendicular to most external measurement sites, i.e. your arteries and pressure goes along your arm rather than out of it. In order for us to measure without sticking a tube in your arm (arterial line), we create an external pressure using a cuff.
The methodology goes as such:
Increase the cuff pressure until blood stops flowing through it. At this point the external pressure is roughly equal to the maximum blood pressure (systolic) because it can’t push through.
You decrease the pressure in the cuff gradually, during this period, not all blood flows through the cuff well and flows in a different way (turbulent flow).
When the external cuff pressure decreases to the point where blood flows through naturally (laminar flow) we call this the diastolic pressure.
We have to create an external change in the body's circumstance in order to be able to see how it reacts, and from that be able to measure blood pressure.
Now there are other ways to measure BP, such as tonometers and the finapres. But these are the least used due to the practical applications as brachial and arterial pressure is the standard for decisions in medical care.
Location, Location, Location
Why is the upper arm currently used?
There is a reason BP cuffs are used on the upper arm (brachial artery), it's because it's the closest artery to the ascending aorta that we can easily get access to (without strangling you). The ascending aorta is the artery connected to the heart where blood is pumped out. We want to know this pressure (known as central pressure), because it indicates how much work the heart is having to do to pump blood and if the pressure is enough to push the blood to the head.
As you get further and further away from the heart, your arteries start to get stiffer and there is a bigger effect from what is called pulse wave reflection. These effects causes your systolic pressure to rise. When you're young, there is a bigger difference in blood pressure between your heart, arm and wrist. As you get older or you get atherosclerosis, and your arteries stiffen, the pressure become more similar but they are not the same. There is a roughly 20-30 mmHg systolic difference between your heart and your finger when young and 0-10 mmHg when older, but this is not the same for everybody and can be very different.
Not only this, the blood pressure waveform (which is collected by the sensors and used to measure blood pressure) can change drastically between your wrist and your heart. What changes is you lose fidelity on what's called reverse wave reflection. I won’t go into it too deeply here but it's a part of the blood pressure waveform at best seen at the the heart and arm which indicates how stiff your main arteries are. Because of the way the waveform changes to the wrist, this information is often lost.
What's wrong with the wrist?
If there was a site on the body that's probably the worst to make any kind of measurement from, I would say it was either the finger or wrist. What smartwatches use to measure BP is called a reflectance PPG sensor. This sensor sends different colour light into your skin, and this gets reflected back out to the watch. These different colours (green, red and infra-red usually), get absorbed differently depending on how much blood is in your vessels. This creates a waveform with the pressure characteristics that you would see from a blood pressure cuff or an arterial line (although they are not the same).
Now here are the big issues with the wrist:
Bone structure: The bone structure of the wrist makes it very difficult to make consistent contact with the skin. The PPG sensor is extremely vulnerable to external light and needs really good, consistent contact with the skin (think using a projector in bright light). The bones on the wrist make it very difficult to do this reliably, although it can be done if not for the second point.
You move around a lot!: When you move, this causes the contact between the skin and the PPG sensor to change. This change makes the data unreadable. There is no AI that can recover this! Any AI that does is just seeing what it thinks is there, not what is really there (estimation, not measurement). Of all of the parts of your body, the finger and wrist move around the most!
Veins: Look down at your wrist, you probably notice a lot of blue lines, your veins. These return deoxygenated blood to your heart. They have different pressure wave characteristics to your arteries and can interfere with reading arteries. This can also cause havoc for SpO2 readings. So placement of the watch needs to be precise, which is impossible and unrealistic to do with a watch.
Overall this means that even if in the best controlled circumstances the watch is accurate, the actual day to day usage of the product will create bad readings.
So how do watch companies do it?
There is a relationship between how stiff your arteries are and how fast your blood pressure wave moves, known as Pulse Wave Velocity (PWV). As your arteries get stiffer your pressure wave will travel faster from your heart to every part of your body. They use what's called Pulse Transit Time (PTT) to measure how fast the pulse wave takes to go from your heart to the wrist. Now there is a relationship between stiffness and increased pressure, which is what PTT is trying to measure. However it is not a direct 1:1 mapping, your stiffness changes based on other factors such as your baroreflex system (if you exercise or feel stressed for example, your arteries will stiffen). There are other variables that can interfere with huge amounts of non-linear interactions.
Could an AI model measure blood pressure at the wrist? I think the question is could it measure this accurately enough for it to be meaningful? I am dubious that enough information in the waveform and PTT is there to truly “measure” blood pressure anywhere near the accuracy where it's clinically (or non-clinically) meaningful.
A change in blood pressure could probably be measured, i.e. it can measure if your blood pressure increases. There are already FDA cleared devices that do this. But they cannot tell what your blood pressure is and again, the accuracy may not be good enough for any clinical relevance.
There is interesting research ongoing to do with ultrasound and other measurement methods, but it is still early days.
I think the real tell would be whether or not the BP algorithm or AI needs your age, height, weight and sex. These are big factors on your blood pressure. These could be used to create an algorithm that will tell you what your blood pressure probably is, but cannot tell you what it actually is.
How do I know they are accurate?
This is really important! There are 3 main standards (tests to prove accuracy) which blood pressure monitors conform to, AAMI (Association for the Advancement of Medical Instrumentation), BIHS (British and Irish Hypertension Society) and ESH (European Society of Hypertension). They are fairly similar in what they are asked to do. You need to check that your blood pressure monitor has passed these standards! If you have a BHS or BHIS cleared device, only use an A/A rated device, do not choose anything else.
Until watches can pass these standards then they cannot be trusted to provide accurate clinical information on blood pressure.
What's in a word?
You remember I said before that companies will use words to make it sounds like it can measure blood pressure. Well the ASUS watch claims to “detect blood pressure”. This means absolutely nothing. They are using the words they can legally without passing any standards, because they have not passed any standards. If you see a product without a link to a medical standard and the word “measurement” it is not useful to you.
Are smart watches useful for healthcare?
Now the Huawei Watch D has claimed adherence to the AAMI standard. It uses a traditional cuff based interface built into the watch, rather than PPG/ECG. However there only seems to be one study for this based in Malaysia in 2022, primarily on Chinese, Malay and Indian participants, so it would be great to see a validation study on western and African populations as well. Despite 3 years on the market in China, it has yet to be FDA cleared which is the world gold standard for medically applicability. Wrist based monitors however are not primarily used in practice because in the real world, there tends to be large discrepancies in readings, despite what validations can show (this is based on anecdotal feedback).
BP as its measured now is useful because of the body of evidence gathered around its use. Its used for medications and clinical decisions. There are other metrics that could be more useful that could come from PPG / wrist base usage. PWV for example, is a potentially great measure for arterial stiffness which could be more beneficial than current BP. Its just that we need to build up the clinical applications of it as a metric to be able to use it to guide care. If the Apple or Asus took this route I think it could provide much better care!
Overall
If you are a patient, healthcare provider, start-up founder or investor; you should still be highly skeptical about these things. From what I’ve seen these devices are nowhere near being clinically useful and any claims like these are used to obfuscate and trick people into believing they have the abilities to measure blood pressure. Is it impossible? absolutely not. But until companies put measurement of blood pressure first, watch second, they are not going to be able to make a break through to create a truly cuff-less blood pressure monitor or something better.
If your a medical device manufacturer or looking to develop AI in healthcare, reach out to me and we can see how I can help.
