Stress Sensing Stickers

In this episode, we discuss the shortcomings of previous attempts at making flexible wearable sensors and how researchers at CalTech have addressed them to create high performance stress sensor stickers.

This podcast is sponsored by Mouser Electronics. 

EPISODE NOTES

(3:12) - Measuring Stress

This episode was brought to you by Mouser, our favorite place to get electronics parts for any project, whether it be a hobby at home or a prototype for work. Click HERE to learn more about printable sensors and why they might be the key to wide scale adoption of wearable health devices!

Transcript

What's going on party people? Welcome back to the NextByte podcast. And in this one, we're talking about a sticker that will literally save your life. I'm serious. It's a little wireless sensor flexible from Caltech that is gonna measure your stress in real time and let you know if you're stressing out. So, if that's got you excited or at least stressed, buckle up and let's get into it.

I'm Daniel, and I'm Farbod. And this is the NextByte Podcast. Every week, we explore interesting and impactful tech and engineering content from Wevolver.com and deliver it to you in bite sized episodes that are easy to understand, regardless of your background. 

Farbod: Alright folks, as we mentioned earlier, we are talking about wearables this episode. And before we jump into today's episode, we're going to talk about today's sponsor, Mouser Electronics. Now, Mouser is one of the world's biggest electronics suppliers, and what that means is they've got great connections to folks in academia doing groundbreaking research and the latest in industry, because again, they're the suppliers. So, they've written this article about the... Sorry, my cat was making a little mess, so I had to stop that. But they've written this article about how there's been this crazy adoption of wearables over the past few years, kind of spurred on by the COVID-19 pandemic, because obviously people wanted to do remote health monitoring, and telemedicine became really big. Wearables like think the WHOOP strap, Apple watches, things like that, that could give immediate feedback on how you're doing. They became popular.

Daniel: Oura Rings.

Farbod: Oura Rings, yeah, yeah.

Daniel: Nellie and I have Oura Rings on the way, by the way.

Farbod: Oh, nice, okay.

Daniel: Little Valentine's Day gift to ourselves.

Farbod: Love that for you, guys. They became super popular. And in this article, they talk about what the next generation of those products look like. Why not move to fully printed, flexible, very thin sensors? And it makes a lot of sense when you think about it because it's much easier to manufacture, uses less materials, it's quite a lot cheaper. And that kind of serves as the basis for today's topic. So, if that type of technical knowledge is of interest to you or that domain is something you find really interesting, I highly suggest checking this article out. We're gonna link in the show notes as always.

Daniel: Well, I just say as a technical primer to today's topic in general, right? They talk about all the different methods for manufacturing printed sensors and all the benefits of using printed sensors. Like you mentioned one of them being really thin and flexible. This team from Caltech that we're talking about today, more boxes falling thanks to Farbod's cat. But the topic we're talking about today, these scientists from Caltech created a special sticker, which they're using for health monitoring. Spoiler alert. But they use at least one of those layers is inkjet printed, which is one of the methods that's mentioned in this Mouser article. Yeah. Very very pertinent even as a primer for what we're talking about today.

Farbod: Absolutely. I totally agree and with that said let's jump into today's article We're going all the way to the west coast. We're gonna be talking about folks over at Caltech and the research they did with again spoiler alert flexible wearable sensors. Now, before we get into it where the topic of the day is gonna be… Kishwish is really on one today. Where was that? Yeah, the topic that we're gonna be talking about today is of course these wearable sensors, but with a specific focus on sensing stress. And what I thought was really interesting that was mentioned in this article is that the term of what stress is was coined by physicist and chemist, sorry not physicist, physician and chemist.

Daniel: I was about to say, someone who's in physics and chemistry has nothing to be doing with their brains.

Farbod: Yeah. You know what, my brain is working slow, Kishwish was just throwing them off, I wanna blame her.

Daniel: Maybe you're stressed.

Farbod: Yeah, maybe I'm stressed. Maybe I am stressed. But it was by Hans Selye and I'm definitely butchering that who was…

Daniel: Austrian, right?

Farbod: Austrian. Yeah, born in 1907, which makes you think like stress hasn't been around that long. Yeah, you know, maybe I should stop stressing about stress but it's funny because the term was coined as a non-specific response of the body to any demand. So, it's pretty generic. It's very non-specific and what's funny is that when you look at stress in general? It is indeed non-specific. There is no one biomarker that you can use to identify that someone is going through a stressful event.

Daniel: Well, and stress manifests differently in different people.

Farbod: Absolutely.

Daniel: One of the things, and they mention it in the article, they say stress is a slippery concept, and I like the word slippery, right? You could be feeling stressed, and those symptoms could be manifesting one way in you. Maybe you've got clammy hands and maybe you start stuttering when you're speaking. But if I'm stressed, maybe I have a stress headache. Or like I start to grind my teeth at night. The symptoms of stress manifest very differently among different people. And there are also similarly, if you were to try and measure stress, there are tons of different biomarkers that you need to measure to be able to understand the full picture of the human body if this person is undergoing a stress event. And one of the things I liked another piece of verbiage they use is they call it a constellation, right? You need to measure all these different points and kind of map their relativity to each other. Kind of like if you're looking at stars in the sky and trying to look for the Big Dipper, you need to map this constellation of different points to try and understand if someone is experiencing a stress event. And one of the things they’ve mentioned is you have to start measuring before they're in a stress event to understand when this constellation changes. So, then you can be like, oh, because there's a change in all these different signals, we can say that this might be a stress event. And there's also a lot of false positives. They had to weed out as well. Like you can't use purely just an increase in amount of sweat as a marker for stress because I might sweat more when I'm working out and maybe I'm actually enjoying that I'm playing pickleball with Nellie, right? I'm a relaxed, I'm happy. I'm enjoying it. I'm actually the opposite of stressed, but if you're measuring just sweat, you'd be like, oh man, he's stressed out right now.

Farbod: And you know what? If folks don't take away anything from this episode, let it be that just like the stars, these constellations, they are proof that there is scientific evidence that astrology has value, it has meaning. Your sign means everything. Let that be your takeaway.

Daniel: I don't know about that, man.

Farbod: But like, just to kind of like carry on what you were saying, absolutely right, stress manifests differently in different people, but almost across the board, it manifests in two ways for everyone. There's the chemical response of your body, like, you know, you have glucose I think in your sweat that spikes, you have cortisol in your blood, you have physiological responses as well, like you mentioned some of them like stuttering and whatnot. But even though it's difficult to characterize stress on a person to per person basis, what we have or what physicians generally agree on is that stress is bad. Like acute chronic stress results in mental and physical damage.

Daniel: Well, that's what I was going to mention here is, I pulled just a couple of this potential long-term effects of chronic stress include shortened lifespan, increased risk of heart disease, weakened immunity, increased rate of aging. It's also mentioned that specifically for people who are in high demand occupations think athletes, first responders, doctors, air traffic controllers, people who, you know, it's really, really important for them to be at their mental and physical best. Those also tend to be subject to elevated levels of stress and anxiety, which interfere with their cognitive performance, their decision-making process, which could ultimately end up in accidents impacting people who aren't even stressed. Think about if you have an air traffic controller who's under chronic stress because of their job, it's very high demand. If it impacts their cognitive performance, impacts their decision-making ability they can make a bad decision that causes two planes to collide and impact hundreds of people who weren't even the ones dealing with the stress in the first place.

Farbod: Yeah. So, like just to recap of where we're at, stress, very hard to identify from a person-to-person basis. Stress, very bad for people, both on a physical and mental level. And stress, especially if you're in a high impact role, very intense role, it can not only damage you, but people that you're supposed to be taken care of and those around you. So, how do we solve this problem? Like how do we get around this or come up with some sort of innovative solution and that's where the folks at Caltech come in. They're like, hey, in the past, we have been able to create wearable skin sensors that pick up on some of those biomarkers that we care about for stress, right? But they just don't, even like the state of the art, it doesn't perform that well. Now why doesn't it perform that well? I'm posing the question, you know, trying to toss a ball to you. So why doesn't it perform well?

Daniel: Well, I think a couple of different things, even before we jump into that, right, is like there were lots of different markers that need to be measured. You need to sample those at a proper rate and you need to process those in tandem to try and understand what's going on. There's a lot of previous research that they had completed, which I think they called it micro sampling or microfluidic sampling, which proved that they could pull tiny bits of, I think it was sweat and run sampling on those to measure some things like nickel-based compounds. But honestly, not sure what was the major roadblock that you were thinking of?

Farbod: No, no, no. You're kind of on the money. So, the biomarkers that you care about, right? You have the stuff that you're getting from your sweat, but it kind of, I guess, breaks into two different groupings. There's the ion-based detection that you want to do so that picks up on like your sodium and potassium levels, and then you have the enzymatic based detectors that are supposed to pick up on your like lactate and glucose. So, the microfluidic approach, if I understand it correctly from the article, it was only able to pick up one grouping of it. But on top of that they were using inkjet printing, which is great because it's very quick, very scalable manufacturing process. But the materials that they were using degrade it actually quite substantially when it encountered the human, liquid's not the word, what am I looking for?

Daniel: Oils, sweat.

Farbod: Yeah, discharge, there we go, human discharge. Like your bodily fluids.

Daniel: I like the liquids more than discharge.

Farbod: Okay, we'll go back to liquids. But it degraded, which meant that the sensor was no longer usable, or like, you didn't have a lot of confidence in the data that you were getting. So, you have this problem of like limited data that you're looking at to analyze and therefore your data isn't that gonna be, the processing is not gonna be that great to begin with, but then the device itself just degrades over time, which means it's not really dependable. So how do they go about solving this problem?

Daniel: Well, there were a bunch of different things that I dove into. The one that I thought was most interesting is just understanding how many different layers they have in this stack, this sensor, that allowed them to be selective to different types of ions, right? So, they've got three different ion selective sensors, three different enzymatic biosensors, sweat stimulating electrodes that are made with this carbacol hydrogel, a capacitive sensor for measuring pulse, a resistive sensor for galvanic skin response, and a skin temperature sensor. All in all, they're measuring nine different stress markers and they're able to pack this all into one thin little patch that is flexible and sits in your skin. But the way that they were able to separate all these different things that they wanna measure is by channeling fluids to different layers of this thing. I don't know honestly how many layers there are in total, but I think it was at least four or five different layers stacked together in this patch with an air gap between certain layers to make sure that the parts of the, let's call it human liquids or the sweat, the discharge from human skin wasn't breaking down certain aspects of it if you're trying to measure more than one measuring end at once.

Farbod: Yeah, yeah, I know you're absolutely on the money. And just for folks to know, I didn't know what galvanic skin response was.

Daniel: Not me either.

Farbod: It's just your skin's electrical properties, basically how conductive it is, as a response to the activity of your sweat glands. So, like as my palms start sweating because I'm getting stressed, it becomes more conductive. So that's like something that they're monitoring. But then there's the other factors that you mentioned, such as the temperature that they're looking at and the waveform. So, the pulse waveform is like you have your blood pulsing through your body that allows them to, I'm pretty sure, decode it and pick up what your heart rate is looking like. So, they have all this different processing that's happening. You have all these nine biomarkers, but how does it address the issue that they were seeing before with the degradation of materials?

Daniel: I think it was they doped it with nickel, added nickel-based compounds that helped, I think that was for the enzymatic sensors.

Farbod: You're right.

Daniel: That helped those from breaking down. They also upgraded a bunch of the polymers for the ion-based sensors, which made them more stable over a long period of time.

Farbod: Yeah, yeah, yeah. So, for folks that don't know, nickel, very corrosion resistant. Inconel is a nickel alloy commonly used, it's a nickel steel alloy actually, commonly used for applications where you want corrosion resistant material. So same use case here, you have your body's sweat being secreted, if we don't want to say discharge, and that it has salt in it and other materials. So, the addition of nickel makes it a lot more stable. And then we have what you were saying much earlier on, which is the constellation of data. Like how do we map everything together? And I think out of their three-part sauce, one part being the whole design of everything, the other part being the new material they're using, then there's this third part that actually brings the entire sauce together, right?

Daniel: Yeah, well, and I'd say the first two would allow them to establish a proof of concept, let's say. These are a bunch of the things that we wanna measure. We were able to pack them all into one physical device that we know sticks on the skin and is able to measure these specific signals properly some way of interpreting the data.

Farbod: It's just meaningless.

Daniel: It truly would be a laboratory only development. It wouldn't be able to detect stress in real time just because of, just imagine the complexity of measuring all these nine different sets of data, all these nine different signals. If you as a person were monitoring that and to try and understand whether a person was stressed or not, the way I like to liken it was like, you had to be like an expert lie detector. You know, the people that are really, really good at reading polygraphs. Like they look at all the different signals and they watch the little, I don't know, the traces, the traces going. You need to know which needles are spiking in a certain direction to be able to detect whether that person's lying or not. I think if you were monitoring stress, you'd have to have the same level of expertise. I have no idea how to do that. And I think even these scientists would be really, really hard to in real time monitor nine different signals at the same time. And if it's hard for the human brain to do it, it's massively harder to be able to design a program to do it computationally. And so, this third part of the secret sauce you're mentioning here, I'm going to tee you up for it. But how do they solve this big problem?

Farbod: Machine learning. Yeah. Right? Like, that's what makes sense. That's been the hype for at least the past two years. And that's exactly the tool that they're using here. What makes that challenge more difficult, by the way, before we move on from your little example of the lie detector, is that this response is also quite different from person to person. So not only do you have some sort of connection being made between these nine variables, but those nine variables differ in level and significance from one sample to another. So that's where machine learning really comes into play here. And that's what they use to kind of try to map how these variables should correlate with each other in a stressful event. Obviously, it's one thing to have that design and it's one thing to say, we made it. It's another thing to say, here's how it actually works and that's where it gets pretty interesting for me because they brought a pool of testers in and they took, they got feedback from them like are you stressed right now, are you not stressed? And they took readings from them using this device then they put them through a stressful event. For example, like a very extreme video game scene or some sort of stressful event that gets a response out of them and then they ask them again afterwards like how are you feeling, how are you feeling now? And there was great parity between what the device was telling them and what the people were telling them.

Daniel: Well, and they mentioned like purely physical stress events or something like putting your body through a really extreme vigorous exercise and a very extreme mental stress event. Those are, you know, we would all, we would consider both of those to be stress, but those are on vastly different ends of the spectrum. So, what they did, and I think you hit the nail on the head there, they did one test at one end of the spectrum, which was like, do vigorous exercise. We're gonna measure your stress signals before and after. And then they also did the same thing, but with like a really, they called it intense video game play. I'd be interested in understanding what exactly that video game scenario was, but-

Farbod: I wish I was a tester, I'm not gonna lie to you.

Daniel: Yeah, I mean, but like, so they took this person from sitting at rest, again, going from zero physical stress, but also zero mental stress, and then cranking up the mental stress aspect, and then measuring how that goes. And by looking at both ends of the spectrums; I think they did a pretty good job at correlating what the different types of stress responses could look like in a person. And honestly, like, we would, I think the colloquial term would call it like calibrating, right? You calibrate this tool to understand this is when I'm not stressed, this is when I'm stressed. This is when I'm stressed physically, this is when I'm stressed mentally. But in the machine learning realm, we call that training, right? You're training the machine learning algorithm to understand different types of stress responses.

Farbod: Get the baseline in, yeah.

Daniel: And I think this becomes more and more interesting the more and more subjects they test with, because I think that they'll probably be able to build a couple of different archetypes. I know that stress responses may look different from person to person, but I bet that there's a couple different categories of people, so that this thing might work really, really well out of the box for someone, as opposed to saying like, here, you have to run through these test events for me to be able to work properly as a stress monitoring tool.

Farbod: Yeah, that's totally a fair point. And in terms of like what this device might look like outside of a laboratory setting, like if you were to buy it or I were to buy it and you know go through the unboxing experience and use it, this thing can be battery powered, so that's nice. And it has Bluetooth on board so it's able to pair with you know a smartphone or whatever it needs to, to transfer data and upload anything that you might want to. So that was really nice to see. I feel like obviously they had those additions to make their testing a little bit easier but it also means that they had that feature set in mind in case they want to commercialize it at some point down the road. But yeah, this is kind of the next stage of again what is now our Apple watches and Oura Rings, and if it comes to the shelves in the next couple years, I will be first in line to buy one. I would love to have one and wear it throughout the day.

Daniel: Well, that's what I was gonna say is you and I have gone on this rant a couple of times together saying like, man, I really, really wish I could get the same fidelity of data that I'm getting from my Apple watch, but I wish I could wear my normal watch alongside with it. Yeah, we're big watch fans and we're aspiring to grow our watch collections, let's say. But for me personally, because of the data I get from my Apple watch, this is the one that I end up dailying every single day, regardless of whatever watches I have on the shelf.

Farbod: And I have to sacrifice my health just to wear the watch I like.

Daniel: But this is an interesting scenario, right? It's a sticker that sticks on, I think they, they said on the inside of the wrist is where they got the best readings. And it looks about the size, if you were to imagine like two quarters right next to each other, that looks about the size of one quarter being the sticker patch and then another quarter being the Bluetooth module and the battery. It seems pretty non-invasive. It seems like I could totally stick that on my non-watch wrist and have no problem with it. Or maybe further up my arm if I was wearing a long sleeve shirt and it could be completely concealed too. It's thin, it's flexible. They showed that it sticks on, it stays on throughout this sweaty exercise. It's flexible so it didn't impede their movements or didn't feel uncomfortable for them. And the subjects that tested with it, they did this vigorous exercise and even then, it was still able to get all the readings and stuff like that. To me, that sounds a lot more interesting maybe than wearing the Apple watch. If it frees me up to be able to wear a watch from our collection or not wear a watch at all, have some freedom of movement and exercise and stuff like that with just this really thin, slim sticker, which I think WHOOP might be the ones that I feel like pioneered kind of this language of treating exercise as physical stress that gets built into your body and then you need to recover from it.

Farbod: They call it strain.

Daniel: Yeah, strain, there you go. Stress strain curve, yeah. But it would be an interesting step forward for us to see here if they didn't just use it for stress monitoring, but they also tried to monitor the inverse, right? Measured times where you weren't stressed to also quantify your recovery as well.

Farbod: Yeah, yeah. So, let's do a little quick recap, right? Little over 100 years ago, this guy who was a chemist and a physician coined the term stress as a non-specific response of the body to any demand. And as the years have gone on, that's still the case. There is no single biomarker that tells you that, hey, this person is stressed. There's a whole variety of chemical and physiological responses that happen, and that varies from person to person. So how do we determine if you, my friend, are stressed or not? Well, these folks at Caltech, they've come up with a great solution that is looking at nine different biomarkers and mapping it together while also using a machine learning model to analyze if you are going through a stressful event or not and then notifying you about it. This thing is completely wireless. It's incredibly thin. It's a flexible little basically band aid that you can wear on your arm throughout the day. It won't come off if you exercise. It can hold the power as you go to bed. And throughout the day, it will let you know if you're going through a stressful event so that you can take a nice calming walk or enjoy the breeze or listen to music or do whatever you need to do to calm down.

Daniel: Money. Perfect.

Farbod: That's it. That's the sticker that saves your life. That is the main takeaway that you should have from this episode.

Daniel: I think obviously we're calling it the sticker that measures stress or the patch that measures stress in your body. I think it can open up to a whole lot more. If you're measuring those nine different biomarkers, I'm sure there's a lot of different aspects of health you can measure a long side stress. Given that stress is such a hard thing to measure, if you're catching that much, you know, that many signals about the body, I can only imagine all the different conditions and situations where this information could also be useful.

Farbod: Yeah, they should go for a strain next.

Daniel: Yeah, how about that?

Farbod: Strain. All right, we good to wrap it up?

Daniel: I think we should, we owe some folks a thank you.

Farbod: Ooh, I forgot, go ahead.

Daniel: We were trending in three different countries. I picked my favorite of the three.

Farbod: All right. Well, you're going to offend the other two, so.

Daniel: Well, we're not going to spill the beans on the other two, so I don't get canceled.

Farbod: It's OK. I'll tell them later. Cancel you.

Daniel: We trended at number 46 podcast in Oman. OK. So, thank you to our friends in Oman. And I have to say thank you, so I'm going to say “Shukran” to our friends in Oman.

Farbod: You chose that because you knew what it was in there before.

Daniel: I've also used that one before.

Farbod: Shame. Taking the easy way out. I'm gonna expose you later.

Daniel: I will also say “Xie xie ni men” to our friends in Taiwan, which is another one that I already learned in the past.

Farbod: Okay, all right, it leaves one more. I guess we'll get to the next one.

Daniel: Yeah, it's hungry, I don't know how to say it.

Farbod: Ooh, all right, yeah, we'll get to the next one. I didn't Google it either. To our friends in Hungary, we apologize. Well, we promise we'll give you a hearty thank you next time.

Daniel: Yeah, we'll catch you on the next one.

Farbod: Awesome, thank you guys, and as always, we'll talk to you later.

Daniel: Peace.

As always, you can find these and other interesting & impactful engineering articles on Wevolver.com.

To learn more about this show, please visit our shows page. By following the page, you will get automatic updates by email when a new show is published. Be sure to give us a follow and review on Apple podcasts, Spotify, and most of your favorite podcast platforms!

--

The Next Byte: We're two engineers on a mission to simplify complex science & technology, making it easy to understand. In each episode of our show, we dive into world-changing tech (such as AI, robotics, 3D printing, IoT, & much more), all while keeping it entertaining & engaging along the way.