Code & Cure
Decoding health in the age of AI
Hosted by an AI researcher and a medical doctor, this podcast unpacks how artificial intelligence and emerging technologies are transforming how we understand, measure, and care for our bodies and minds.
Each episode unpacks a real-world topic to ask not just what’s new, but what’s true—and what’s at stake as healthcare becomes increasingly data-driven.
If you're curious about how health tech really works—and what it means for your body, your choices, and your future—this podcast is for you.
We’re here to explore ideas—not to diagnose or treat. This podcast doesn’t provide medical advice.
Code & Cure
#8 - No Cuff, No Problem? The Future of Blood Pressure Monitoring
What if checking your blood pressure was as easy as glancing at your watch? High blood pressure quietly affects nearly half of all Americans—yet it's one of the most preventable causes of strokes, heart attacks, and other serious health problems. The catch? Traditional monitoring methods are clunky, inconvenient, and rarely used outside the clinic.
In this episode, we explore how next-gen technologies are transforming blood pressure tracking. From smartwatches and rings to toilet seats and even facial recognition, wearable devices are pushing the boundaries of what's possible—no cuffs required. You’ll learn how sensors using light (PPG), electrical signals, and video can estimate blood pressure in real time, offering the promise of continuous, hassle-free monitoring.
But as with any innovation, there are hurdles. We dive into critical challenges like calibration complexity, variable accuracy across users and activities, and whether these tools truly improve hypertension management or simply add more data noise. The role of artificial intelligence adds another layer—enhancing insights, but also raising new questions about equity, access, and interpretation.
Is convenience enough to spark a shift in how we manage cardiovascular health? Or do these tools need to prove more than novelty to become essential?
Tune in for a forward-looking conversation on the promise, the pitfalls, and the future of blood pressure technology—where innovation meets one of medicine’s most familiar numbers.
References:
Emerging sensing and modeling technologies for wearable and cuffless blood pressure monitoring
Lei Zhao, Cunman Liang, Yan Huang, Guodong Zhou, Yiqun Xiao, Nan Ji, Yuan‑Ting Zhang, Ni Zhao et al.
Nature, Digital Medicine, May 2023
Cuffless Blood Pressure Measurement Devices – International Perspectives on Accuracy and Clinical Use: A Narrative Review
Eugene Yang, Aletta E. Schutte, George Stergiou, Fernando Stuardo Wyss, Yvonne Commodore‑Mensah, Augustine Odili, Ian Kronish, Hae‑Young Lee, Daichi Shimbo
JAMA Cardiology, June 2025
Credits:
Theme music: Nowhere Land, Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0
https://creativecommons.org/licenses/by/4.0/
Say this five times fast. Optical photoplethysmography.
Speaker 2:Welcome back to Code and Cure. My name is Vasant Sarathy and I'm here with Laura Hagopian. Hey, laura, how are you Good? Good, how are you doing?
Speaker 1:I'm excited because today's topic of blood pressure is one of my favorites.
Speaker 2:The unpronounceable.
Speaker 1:I know I gave you a challenge and you didn't even try.
Speaker 2:Oh, my God, what did you say? Photo.
Speaker 1:Optical photoplethysmography.
Speaker 2:Optical photoplethysmography. Is there an acronym that we could use instead? Maybe?
Speaker 1:I mean we could call it PPG.
Speaker 2:Yeah, ppg. Okay, I like that, I like PPG. But I'm also very excited about this topic and it has to do with monitoring blood pressure and doing that in a really smart way. That can actually be really powerful.
Speaker 1:Maybe, maybe.
Speaker 2:Maybe, yeah, we're going to get ahead of ourselves a little bit. Maybe you know this is the tech nerd in me that's excited about applying technology to everything. But yeah, let's talk a little bit about blood pressure itself.
Speaker 1:Yeah, so blood pressure is basically the pressure of blood pushing against your arteries, right, and it's reported as two numbers, so you might hear something like 118 over 76. And the 118, that first number is called your systolic blood pressure. That is the pressure in your arteries when your heart is beating, so when your heart is like pushing blood out into your body. And then the 76, the second number is the diastolic, and that is the pressure in your blood vessels and your arteries when your heart muscle is resting, so that's like when not when your heart is pushing the blood out, but when your heart's actually filling itself with blood.
Speaker 2:Okay, so is it in between then, there's a pushing out step and there's a pulling, staying. I mean, is it like a? Then there's a pushing out step and there's a pulling and staying. I mean, is it like a? What do you mean? It's not the pulling out, it's okay, it's the heart is not pushing blood out.
Speaker 1:Instead, you know there has to be.
Speaker 2:The heart has to fill with blood and then the heart pushes the blood out to the rest of your body right it's obviously more complicated now.
Speaker 1:The blood has to go to the lungs and get filled with oxygen, exchange for carbon dioxide, etc. Etc. Etc. But but what we're talking about?
Speaker 2:those two numbers are like the highest and kind of the lowest blood pressure now that your body's seeing when we talk about those two numbers are we talking about? Can you relate that to? Like the heartbeat because that's kind of where my starting point is with the heart is always the boop boop, boop, boop.
Speaker 1:You know the yeah, I mean that first number is really when your heart is beating, pushing the blood out, and then the second number is when your heart is relaxing and filling with blood. Gotcha gotcha, and blood pressure is important. Right Half of Americans have high blood pressure, and most people who have high blood pressure don't have it under control, which can lead to lots of problems down the line.
Speaker 2:Yeah, let's talk about that. What kinds of issues are sort of based on high blood pressure?
Speaker 1:Yeah, I mean it damages your blood vessels, right. So anything anywhere where the blood has to reach to, right, it can cause problems in your brain strokes, right. It can cause problems in your heart strokes, right. It can cause problems in your heart, heart attacks, heart failure. It can cause problems in your blood vessels themselves where you're not getting enough blood and you get something called peripheral vascular disease. So there are a lot of complications that can come from high blood pressure that's not under control. So then of course, it's important to check it. It's important to measure it. For people who have potentially have, or who do have, high blood pressure, it's important to track it to make sure hey, do I actually have this condition right and do I just have it the one time my doctor checked me, or do I have it all the time? And then, if you're on treatment, is it actually working? Is the medicine or the lifestyle changes making the difference that we want to see? Because if you have high blood pressure, we want it to be down below a certain number.
Speaker 2:Yeah, and to your point about the one-time measurement versus all the time, you know it's one of those things when you go to the doctor's office they ask you to like think happy thoughts when they're measuring blood and measuring your blood pressure, because they only have one or two chances to do it when you're there and so you have to hope that that is representative.
Speaker 1:It's interesting to me because what we're talking about today is measuring your blood pressure without a cuff. But to me traditionally right. You measure your blood pressure directly with a cuff and that checks those two numbers, the systolic and diastolic. It directly checks how much pressure there is of blood pushing against your arteries and it's done on the upper arm. We've all had it done, you know. To me it seems like something that's simple and easy. Someone would come into the ER and that's one of their vital signs, along with heart rate, you know, temperature, respiratory rate, all that good stuff.
Speaker 2:Yeah, and frankly, to me as a lay person, I always wondered about this. So you know it makes sense for me to monitor my you know, monitor various signals, right, various vital signs. Blood pressure has always been one that has been a little bit confusing and this might not be true of every lay person, but definitely me which is that I always I never fully appreciated why I as a person would want to monitor it. I get it that the doctors, among the many things that they have to monitor, that blood pressure is one of the things that they need to keep track of, and I get that part. But I get measuring heart rate, for example, and your heartbeats and heart rate and so on. To some degree it shows proper functioning of your body.
Speaker 2:I understand all these Fitbits and so on that have other sensors, like your sleep sensor. That makes sense to me. You want to track how well you sleep Steps. That makes also makes sense to me because that in that instance you're checking how active you are, and that also makes sense. But blood pressure has always been kind of a very medical thing. You do that when you go to the doctor's office, not something that I'm seeking out actively myself, or maybe I should be, maybe that is important, and I think that, as a lay person, that to me I mean granted, of course, if you have issues and you know that you have issues and you're tracking it, that's a different story. But if you don't necessarily know that you have issues on that front, then is that something that it's not something that I would have thought of to track at all.
Speaker 1:Yeah, it's an interesting question because it's something that we ask people to track outside of the hospital, right, we screen people for high blood pressure, right. But then we ask people who have a high number in the clinic to go and check their blood pressure outside of clinic, right? Does it remain high if you check it twice a day? Does it stay at those high numbers, or did you just have a single high reading outside? I'm sorry, inside the office and you don't have high readings outside. So this is a metric that I think, yes, you can track. Half of Americans have hypertension and at some point, they're going to need to track it to confirm their diagnosis and to see if treatment is working as intended. Right, right, well, blood pressure, and I'm just going to track it on your own. You're reaching an understanding with your provider in terms of what these numbers mean, what to do as far as next steps if your numbers come back really high or really low, and setting a goal in conjunction with them. But it's extra information that you could potentially use together.
Speaker 2:Yes, yes, and to me the other piece that's always been baffling not baffling necessarily is the right word, but you have the cuff and you have a way to actually measure it, and that seems quite a big amount of work to be done as separate from all the other sensors that exist right now that you can just put on your wristwatch, right. And so that's why one of the reasons this topic was particularly interesting to me to explore how we can measure blood pressure in a cuffless manner, where it's kind of just part of your regular day.
Speaker 1:But like, is it that hard? This is what was interesting to me about this paper, because there's definitely some novelty to be like. Oh hey, you know you already wear a wearable or like take a picture of your face and you can tell what your blood pressure might be. But like, is it that hard to put a blood pressure cuff on and check someone's blood pressure?
Speaker 1:To me it's actually pretty simple. You know, if someone was coming into the ER, they're getting that vital sign done upon arrival. They're having it continuously monitored, potentially if they're in a telemetry upon arrival. They're having it continuously monitored, potentially if they're in a telemetry or monitored bed in the ER. So I don't think taking someone's blood pressure is hard and I know that with validated tools it's accurate. And so when we look at these cuffless measures, there's definitely like a novelty factor here, like, oh, my toilet seat can check my blood pressure if I wanted to. But I really do have this question of like, you know, what is the potential additional utility here? And I do have a clinical perspective on that for sure, because many people, you know they don't place their cuff in the correct location.
Speaker 1:So if you're checking your blood pressure at home and you, you know, don't put it in the correct spot on your upper arm or you may not check your blood pressure correctly because you don't have the right size cuff compared to your arm and there's definitely like a convenience factor, right? I don't think it's hard to check blood pressure. I do it all the time, but for some people, you know, taking out that cuff, putting it on, pressing, the button, it's another layer of friction and you know people might prefer I could reflect this back to you.
Speaker 1:would you prefer to like have your wearable device check your blood pressure rather than putting on a cuff?
Speaker 2:Yes, I would, of course realizing that there's a potential for inaccuracy, and if our sort of gold standard for measurement of blood pressure is a cuff, then any other system is making some kind of trade-off maybe, and the trade-off is convenience and continuous measurements and those kinds of things. But the trade-off is that what you get with that is potentially lowered accuracy, because you're doing it in a way at least from what I'm reading so far, it seems like what you're doing with blood pressure estimation and all of that in these cuffless settings is you're estimating it. You're not actually calculating the blood pressure itself.
Speaker 1:Right. So with the cuff you've got a direct blood pressure measurement.
Speaker 1:So, with the cuff, you've got a direct blood pressure measurement. But when you use these other techniques the wrist devices, the patches, the temporary tattoos, the glasses, the rings I wasn't kidding about the toilet seat Most of them use this optical photoplethysmography, ppg, and that is a light sensor that looks at the waveform in the microvasculature and extracts information from there to estimate the blood pressure value. So when you use a cuff, you're directly measuring the blood pressure Great, and we have you know ways that we validate tools for that. When you use wearable devices, it's indirect and then you have to estimate the value, and so, yeah, I think a way to think about this, for me at least has been it's you know, ppg is sort of the optical.
Speaker 2:You know, people have these little pulse oximeters they put on their, on their fingers, um to measure oxygen saturation levels in in your, in your blood, and it's sort of the same idea. You send a light, you know, light beam through it and you, you, you track, what, track, what that is. You track the volume of blood in there, which gives you an estimate, potentially, of the blood pressure. And of course, there's other techniques, right, there's electrical methods, there is, I think there's, even ultrasound methods that can do similar things. There's also other mechanical things.
Speaker 2:I was reading about one that's called TAG. I'm going to say this wrong, but it's what is it called? It's TAG and it's like a mechanical approach where they put these again, these pressure sensors, to look for pressure, tiny pressure changes in your blood flow. And so my point is there's a range of other techniques that people are developing that cover the various aspects, from mechanical to electrical, to optical and so on, ultrasound and so on. We kind of focused in a little bit more on PPG because that's getting more popular, right? Is that a way to describe it?
Speaker 1:more research on it. But you know there are other techniques, like, like you mentioned, you know, pressing your finger on um, an oscillometer, or actually there's ones where you can look at a camera and it can do sort of facial video processing wow um. So there's a lot of options for how you could do this, but I think one key theme is that they all estimate the values and that they all require calibration.
Speaker 2:So talk to me a little bit about that. So my understanding of calibration has to do with okay, let me line my estimation up to make sure that it's actually measuring the blood pressure, and the way to do that is I would need some kind of you know, computational people call this ground truth. I would need the correct answer for what a blood pressure should be, and then I can adjust my cuffless sensor to make sure that it's also reading the same thing.
Speaker 1:Yeah, you've got it Exactly. So you have to know. You'd have to say, okay, like what does my blood pressure cuff show at the same time as I'm using my cuffless you know cuffless monitor, or at the same time as I'm checking with this cuffless sensor, and then you know, over time you'd want to make sure that those don't um, uh, don't change, or that you can recalibrate it if needed, and so you don't totally get rid of blood pressure cuffs in this process.
Speaker 2:You're just like translating an electrical signal or some other optical signal into a blood pressure level. That translation needs to be calibrated is what you're getting at.
Speaker 1:Yeah, exactly yeah, and so there's a lot of potential with these devices, like you said. I mean, they're cool, right, yeah, and they're easy to use, they're easy to implement. But we want to make sure that when we check someone's blood pressure and take action on it, right, you might prescribe someone a med or deprescribe someone a med, prescribe a dietary or a lifestyle change. You might make a diagnosis of hypertension based on this. You want to make sure that those blood pressures are valid, and so you'd have to be able to check things like okay, is it working after calibration? How often does it need recalibration? Does it work after exercise, for example? Or how does it work after medication? How does it work when I'm awake versus when I'm asleep? There are all these different scenarios where it's like, hmm, we'd have to check if the calibration is working correctly.
Speaker 2:Wow, yeah, that's huge because in all of those scenarios it could influence the reading and make it so that what you're measuring with the electrical signals is not fully capturing what's actually happening with the blood pressure for that individual in that context, in that situation.
Speaker 1:Exactly. I mean, we expect blood pressure to vary through the day. When we check up blood pressure in clinic or when we tell people to check up blood pressure at home, we're looking for our resting blood pressure. So what we have people do ideally is like sit down and rest in a chair for five minutes. I know that doesn't ever happen at your doctor's office. You know your arm should be at heart level. You shouldn't have exercised in the last 30 minutes. You shouldn't have smoked in the last 30 minutes. You know all of this stuff where you're sort of in this like relaxed state and that's when we measure your blood pressure. But when you exercise or when you have pain, or when you're awake versus asleep, like, there will be variation in blood pressure and you want these sensors to be able to pick that up.
Speaker 2:Yeah, and you know there's also other factors, right, for instance the patient's sort of genetic background and the size of their blood vessels and all of those things, I'm sure. And the you know, structure of the blood vessels probably plays a role right in it as well.
Speaker 1:So this is where I kind of turn to you and I say, hey, like, in addition to calibration, I know that for some of these things they're using machine learning models to take into account more than just, you know, the waveform, for example, that's being seen in order to kind of get at what's a more accurate blood pressure. They're taking into account things like race or medical history or whatever it is, to try to get a better estimation. Because that's what this is an estimation of what the blood pressure might be.
Speaker 2:Yes, yes, yes, and those are very new techniques. I mean, this is all very new stuff, right? And so, while people have been developing machine learning models for blood pressure monitoring and blood pressure estimation, this idea of incorporating other factors is fairly new and very difficult to do. You can get all the data, you can get all the clinical data, you can get all the patient's history and their demographics and so on, but then how do you appropriately bring it into the machine learning models? And different machine learning models require different inputs and are better suited for certain inputs.
Speaker 2:You know there's obviously we've talked about this on the podcast before about how different there's pros and cons to different machine learning models, like linear regression versus deep neural networks.
Speaker 2:Deep neural networks you could sort of give it all the data and then just have it figure out what it needs to figure out to make the prediction, but then it's a huge black box and you don't know what it did.
Speaker 2:You don't know that and that's a huge problem, because if it's taking a factor that wasn't that important to your blood pressure and, because of the nature of the data set, was amplifying the impact of that influence of that, then you're going to get inaccurate readings which won't show up till much later, like the model might do really well in testing because it works on that data set, but when you put it out in the real world it may not do as well If in fact, there's issues of, for instance, a feature that was like race, that was unnecessarily or unduly influential in this, you know. So there's a lot of very interesting angles here. We'll have to see how you know, how these different models shake out. But for sure, the different aspects of these different data points I think that's what researchers are working on is trying to get. How can we feed these into the model so that they learn what they need to learn and the correct thing?
Speaker 1:right because, because it's an indirect measurement, it's like what else can we take into account to get a better? To get get get a better?
Speaker 2:more true reading on things, yeah and and it's not just from my understanding there's not just blood pressure reading like one-time readings, but there's techniques to actually create the whole blood pressure waveform right, and so that encodes so much more in it in a real setting that if we're going to have to make one of those waveforms just from these indirect signals, then we need all the indirect signals to make that accurate, and I think that that's one of the biggest challenges here in this field as well.
Speaker 1:Yeah. And then there's the question of like, who's interpreting all this data? If now you know, if you used to get, for example, if you had a high blood pressure in clinic and I sent you home and I said, take your blood pressure in the morning and evening and record it and come back to me with two weeks worth of data, now you're going to come back to me with 28 data points, right, it's a whole different world if you're having your wearable track it, I don't know all the time, right, You're coming back with a ton of data points. And then it's like well, who's going to interpret all of those? Or is there some way to summarize them over time? Oh, this happened while awake. This happened while asleep, this happened during exercise. Here's what the resting blood pressure was. There's so much data to sift through.
Speaker 2:Yeah, and frankly, that might be another AI machine learning type problem. Right, that might be another opportunity for using a tool like AI to help you make sense of that data, but then, of course, you're relying more on that and again, those accuracy and the calibration and the fairness and the bias and all of that has to be tested in that setting as well, where you're summarizing this information. So there's a separate challenge, but I mean, I think these are technological challenges that are not entirely insurmountable. I mean, there are definitely. It's definitely there's a path to it, and the question is is this something we should be doing? And it seems like we should, because there's value in reducing friction, making it easier for people to measure, and if all the appropriate calibrations and the summarizations happen correctly, then it's potentially of high value, right, right.
Speaker 1:There is a practical consideration I want to bring up, though, too, which is like this concept of the digital divide. You know, if you go to your doctor's office to get your blood pressure checked, they're not charging you specifically for that, they're charging you for the visit. If you go to like a pharmacy, you can check your blood pressure for free at many of them. If you do get a blood pressure device, you can find ones for pretty cheap, right, 30 bucks, or, if you want it to talk to your phone, maybe 50 or $100. Some of these devices that are wearable devices are much more expensive. These rings, even the wrist sensors, some of these things especially if they're going to do more advanced stuff like blood pressure, they become very expensive. And so now you have this whole population that might need to have their blood pressure checked, but they can't use some of this new technology because they're priced out of it.
Speaker 2:Right right right right. I mean the prices do come down over time as these technologies become more common. But if it's creating a systemic unfairness because the folks who actually need it can't actually afford it, then we're not really solving the problem, right. But I mean that's true of every piece of technology, though I would argue. I mean this digital divide problem is true across the board, right.
Speaker 1:And thankfully there are other ways to measure blood pressure right, like you can have a cuff we know that that is accurate. There are validated cuffs out there or you can, you know, go to your local pharmacy and get it checked, for free in most cases. So there are definitely alternatives, but I think it's worth considering when we talk about advancing technology like this, like at what cost?
Speaker 2:Yeah, and I do think the point you made earlier about you know what actually is the value in cuffless monitoring is, I think, still a very fair question. I mean, you read academic papers about this and, of course, there's an introduction section that talks about the value and talks about the overall issue of hypertension and keeping track. But the real question I think what seems to be the case is that do we really need to be continuously monitoring blood pressure in this manner, like, is there a need, over and above, from time to time using a cuff, to like actually just tracking it nonstop?
Speaker 1:Yeah, and there are definitely use cases for that. Yeah, there are. So I think we'll end with a few key takeaways here. Yeah, measuring blood pressure is important. It's important for screening, it's important for people who have a diagnosis of hypertension, it's important for tracking things over time improvement you know whether or not you respond to treatments, et cetera. Cuffless devices could be very useful, but they need validation and they need calibration, and we're not 100% there yet. And AI models like ML can help with this, but I'm not sure they're quite ready for prime time. But it is a technology that I think could make a difference for a lot of people in terms of ease of use, and there's a little bit of fun factor here too.
Speaker 2:Yeah, I mean it's kind of cool to say that my watch or whatever is tracking.
Speaker 1:Tracking my blood pressure for me, yeah yeah, yeah, for sure, awesome.
Speaker 2:Great Thanks for joining us.
Speaker 1:We'll see you next time on Code and Cure.
