Alicia Chong Rodriguez is keeping wearable AI tech close to the chest

0:00Starting a business comes with its share of ups and downs, which is why staying true to your vision is essential. a non-negotiable for Romeo and Milka Bregali, Capital One business customers and co-owners of Ra's plant-based restaurant in New York. Romeo and Milka took a leap of faith when starting their own restaurant, gutting an empty space and building it from the ground up. Every pipe, every wall, every detail. But building from scratch came with a heavy financial burden, which is when they turned to their Capital One business card. With the flexibility of the card's no preset spending limit, they were able to spend more and earn more rewards while bringing their vision to life.

0:36Today, Raz's success is proof that with passion and the right support, it's possible to make your dreams a reality. Learn more at CapitalOne.com slash business cards.

0:51Amelia Bloomer. She was a dress reform advocate from the 1800s who transformed the way that women were wearing clothes at that time. because if you think about the 1800s, women were wearing corsets that would basically be so tight that they would deform her organs and they would be dying at early ages of her life because of this. So Amelia Bloomer led this movement against those damaging corsets towards more comfortable and useful garments for women. and that's why the first pants that women ever wore were the bloomers. Amelia Bloomer is one of the many women who inspire Alicia Chong Rodriguez.

1:42In fact, Alicia named her company Bloomer Tech in honor of the early garment innovator. And like Amelia Bloomer, Alicia is making clothing that can transform women's health. Bloomer Tech is focused on wearable technology. That includes a bra that can monitor and detect women's cardiovascular health, preventing future cardiac issues. Since our inception, we were very focused on making a difference for heart disease in women, right? Like this is an area that has been understudied, underresearched, underrecognized in so many ways. like a lot of these diseases lead to the chronic condition of heart failure or even worse, sudden cardiac arrest, stroke, or heart attack.

2:29Bloomer Tech is on a mission to stop this from happening. This is the kind of life-saving innovation that drives my work as an AI scientist and investor. And I'm so excited to share my conversation with Alicia today. We covered a lot from algorithmic bias and the challenges of developing wearable technology to how AI can save our humanity. I'm Rana El-Khalyubi, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

3:11Hi, Alicia. Welcome to Pioneers of AI. It is so great to have you on the show. Yeah, I'm so happy to be here. So first of all, I want to give a shout out to Dr. Joy Bulamwini because she was our very first guest on the show and she recommended that we interview you. And so I'm so excited that we're finally making this happen. That's so cool. Thank you so much. The three of us actually share the MIT connection. And also the three of us are also World Economic Forum Young Global Leaders. But I think you're the newest YGL. How's your experience been so far? It's been amazing. Meeting so many people worldwide has been an amazing opportunity.

3:49Okay, so today we're going to talk about bras. Yes. So you are the co-founder of a company called Bloomer Tech, which basically builds smart bras for women's cardiovascular health. But before we dive into all of that, I would love to hear more about your personal journey, your experience at MIT, and how all that culminated to your starting Bloomer Tech. To share a little bit about my experience, I came to MIT back in 2015, and it was like a huge accomplishment for me because being from Costa Rica, having done my undergrad in Mexico, it was my dream to be able to come to MIT. So I had dreamed about it for a long time.

4:33And when I was here, I got access to so many researchers, right, like world-class view. And I was interested in sex differences in healthcare. So I talked to many of the labs before deciding which lab to go after my thesis. My background is in electrical engineering and computer science. and the lab that I felt more connected with was the Computational Cardiovascular Research Group because they understood what I wanted to work on in terms of really understanding how sex differences will impact AI and future of machine learning algorithms back then. And I think that led to a whole array of things doing that work that led us to realize that we could build a company.

5:20Yeah. You also, I guess, have a very personal connection to starting the company, and I guess it's your grandmother. Can you share a little bit about that story? So right before coming to MIT, and this is why I focused on cardiovascular so much, I was talking with a group of friends, and we all had a relative that had had a breast cancer in the passed. And all of our relatives fortunately survived. They got diagnosed on time. And we were all talking about it and we're thinking, what's the biggest threat to women's health right now? And that's when for the first time I learned about heart diseases being a threat to women.

6:02And I was curious with my family because my sister had had breast cancer. Now my mom has had breast cancer, but we never talk about heart disease. Who in our family has had heart disease? And I'm the youngest of six and I'm the grandchild that is called after my grandma. So I've always been very proud of my grandma because she was a pioneer of her time. Like she graduated as an OBGYN in a time when women were rarely allowed to obtain medical degrees. So I always looked up to her when I went to a class of engineering and I was one or two women in a class of 40. This is what Alicia's do, you know, because I'm named after my grandma.

6:40I'm getting goosebumps. That's amazing. So for me, it was shocking when I asked my mom, hey, has someone in our family had heart disease? And she was like, oh, you know, your grandma died from a heart attack. That's what she died when I know she died when I was 12 years old. I just never knew it was from heart disease. And then talking to other people, they have stories like that. It's a story more common than we can imagine that it suddenly takes a loved one away. Yeah. You know, one of the things that I'm really passionate about is women's health. And in particular, how can we combine this trifecta of sensors, data, and then AI, both predictive AI, but also generative AI, to really understand and get to kind of the underlying causes of and also diagnosis of women's health.

7:26And I just want to share some statistics. So 44 % of women in the U.S. have some form of heart disease, yet women have a harder time getting diagnosed than men. And to your point, we don't even talk about it. And in fact, women are twice as likely to die after heart failure than men. This is very upsetting, obviously, and you have a kind of a personal experience with this too. How has this gap inspired your work? Yeah. And when you add to those statistics across any age, right, because sometimes people hear this statistic of how women are having worse outcomes and they think, oh, they get heart disease when they're older.

8:04but it's across any life stage that it happens. So when you look at the research that has been done in sex differences, you realize that it has not been until very recent that there's been accumulation of evidence in terms of what's really going on between men and women. And it's because women were excluded from clinical trials for a long time, especially from cardiovascular research. It was only until 1993 that the NIH mandated inclusion in their clinical studies of women. That is crazy that before 1993, yeah, the inclusion of women in these clinical trials was overlooked, right? Yeah. So we've never had that much data from women in general, right?

8:47Like I think that that can accumulate and perpetuate a lot of problems that we see in the field, but there is now evidence on sex differences, which is what's exciting. In the last 30 years, like pioneers that understood early on that women's hearts, just because they're smaller, doesn't mean that they're just smaller size hearts of men. You know, like they're not just little men, how some doctors call it. And it's only women, it's not only bikini medicine, right? Like talking about reproductive health, that like women's health is way more than that. And I think that has led to an opening of an opportunity where AI could either perpetuate problems or fix them if we move forward and collect more data from women that is missing, because that's the reality.

9:37Women's data is missing in many areas. Yeah, and for folks who are listening to us who are not immersed in the AI and machine learning space, how does this data and potentially algorithmic bias become quite hurtful? Yeah, I think the underdiagnosis that we see today, right, like if you think about it from a clinical standpoint, we've been using a lot of data sets for clinical decision making for a long time. Either people are at high risk, middle risk or low risk, and that way you decide what therapy they're going to get and what's the next stage for their process. In some of the cases, for example, the CHADVA score, which is a score to measure the risk of cryptogenic stroke, that one has being male or female as one of the things to calculate whether you are at higher risk.

10:27So just being a female puts you at higher risk. Interesting. But that's how little we know about these conditions. They had to put it as part of the calculation. And presumably, the more data we have, the better these algorithms and these decision trees become. So you are focused on a particular piece of clothing, which is the bra. Yes. Why go with the bra design? For us, we write like following her legacy in this technological era. It was important to think about how clothing can do so much more for women's health. And the bra is a garment that women wear every day of their lives when they go outside their houses in many cultures.

11:12So we saw it as something that if we were able to collect data, because it's also in a very good location of the body to collect very reliable and important signals from the body, right? Like we have the heart and we have the lungs right there where we have the bra and opposed to other wearable devices that are making their technology smaller and have to sacrifice a lot of things for the position to make it comfortable, right? Like let's think about a ring or a wristband. The trade-off is their performance and capabilities. We don't have to do that with a bra. We just need to ensure that it looks and feels like another bra, that it's comfortable, but you can use flexible technology to make it seamless experience.

11:58And there's a lot of real estate in that chest area. So you can collect huge amounts of signals. So I want to use this framework of this trifecta of sensors, data, and AI to kind of really understand how Bloomer Tech and how your smart bras work. So let's start with the sensing technology. What kind of sensors do you have in there and what are you measuring? Yeah, so we have our patented textile-based sensors that collect huge amounts of data from that location of the body. So we've put data like the multi-lead ECG. We have temperature, activity, posture, movement. Take us through, like, what exactly is ECG measuring?

12:40Yes. So ECG measures the difference of potential between two points of your body. So it's an electrical measurement of how your heart is working at different angles, right? So there's hundreds of years of evidence how different angles you can interpret the data from the ECG. And now there's more coming thanks to AI in terms of how it can be used for interpreting other things. So it's a very valuable parameter. And you can see all of the different angles of the heart. And right like your atrium, your ventricle, to see how it performs from an electrical standpoint. Amazing. So that's one. And then you said temperature?

13:22Yes. So temperature, we have it in the right location. Also, for many years, we've taken ovulation tracking right under the armpit so that you can get those trends. But it also can give you information about your metabolism. It can give you different information in terms of different inflammation parameters if you correlate it with the right data. But this is talking more bigger picture multimodal. But how is Bloomer Tech's bra different than the array of other wearables out there? We'll get to that after a short break. Stay with us.

14:09If you've spent any time building AI products or leading technical teams, you know this. Transformation doesn't fail because of ideas. It fails because teams can't move together. Enter Atlassian's teamwork collection. It has planning in JIRA, documentation in Confluence, video updates in Loom, and now AI agents in Rovo, which connects the dots across your work so nothing gets lost. It's one AI-powered teamwork platform designed for how modern teams actually build. Learn more at atlassian.com slash teamchanger. That's A-T-L-A-S-S-I-A-N dot com slash teamchanger. so i wear a whoop i wear a sensor and you you kind of started to allude to that um how is the data collected from bloomer tech different than just say a whoop or an aura ring i think it's very complementary but we're just using it for different applications because the physiological signals that we're getting from our device are meant to be used for medical grade capabilities so the resolution of the signal is higher because of the location, right?

15:27Like when you have sensors too close to each other, the measurement is really of that point of the body. When you have them a little bit farther away, you can get better resolution of what you're reading. So it's complementary because our goal is to go more into the diagnostics and clinical side for women that are at risk or have already been diagnosed with heart disease. What are some of the limitations of what this bra can collect today? For us, we've designed it in a way where we can collect data in a dynamic format and the limitations comes with the battery, right? Like the higher sampling rates, the more you reduce the battery capabilities.

16:06So finding that sweet spot of sampling data that you want to get, that will get you to the algorithm that you want to develop, I think those trade-offs are the ones that you have to make good decisions on. Yeah, because presumably you want to have longitudinal data. You're tracking this data over the course of the day. Exactly. But for example, for ovulation tracking, do you need to track it 24-7? Probably you don't need that huge of a sample, right? For other things, you might want to have continuous data. So we also have bioimpedance, which the bioimpedance is a very interesting sensor because it allows you to look at all of the layers of your body in terms of fluid, like sweat, all the way inwards to look at fluid retention in your lungs.

16:52So there's a lot of information there too. So that is interesting. When I first started at MIT and we started our company, we were measuring skin conductance as a measure of arousal off the wrist. And we experimented also measuring it off the sternum. Is this the same kind of signal that you are measuring as well? Yes, I think that one was galvanic skin response. It's similar. It's very similar. Like the methodology of acquiring the data is definitely the same. It's more of like how much you're measuring by inputting different currents and then how you're doing the functions at the end to make the measurement.

17:34Very fascinating. So I want to double click on some of the design considerations to build this into a bra. And I guess the first thing you have to kind of figure out how to do is how do you build flexible and washable circuits? That must have not been easy. I'm guessing this is part of your patented technology. Can you say more about what that looks like? So the interesting part of that is a combination of materials and being very clever on how you connect different technologies. So before coming to MIT, I worked for six years in the semiconductor industry. I think that gave a little bit of insight into what you can and cannot do and how to ensure that you can combine different materials to ensure that this works optimally.

18:21I think the flexible circuit revolution was also happening at that moment where it was finally lower cost and more accessible to make at scale. And that was a very big thing. And then our sensor stack, the way that we build those sensors with textile fabric and ensuring that it was a combination of semiconductors and fabrics so that it could acquire good amounts of data. But we had a lot of combinations, a lot of ideas, and then we just saw what performed better. So if you're measuring things like bioimpedance, for example, how does sweating and exercise and movement come into play? Like that must be another design consideration.

19:02Yes, definitely. I will say for things like electrocardiogram, sweat and all of these things, when it's a dry sensor, it just makes the signal much better and less noisy. So it's actually pretty good when we find people that are sweaty. It's definitely better than having dry skin in that case. And then how do you design for things like varying sizes and even color, right? Like, you know, I don't know if you're going to compete with Victoria's Secret. How do you, yeah, how do you make this both comfortable and, yeah, like the woman would want to wear it? Even though Victoria's Secret is a huge market, we would prefer to partner with them than to compete with them.

19:46Yeah, so definitely the way that we designed it is to be more of a medical device than to be a consumer device, right? Like our application pipeline is focused on clinical decision making rather than health and wellness. So the way that we designed it is to be much better than the current patch monitors, those stickers, or be better than halter monitors, right? Something that can be comfortable, that she can wear every day of her session because she usually has a doctor following her session. And that it won't interfere with her day-to-day activities, right? Like she can wear it as she would any other bra that she's ever seen.

20:27And over time in these design considerations that you're asking, we learned a lot about different preferences in terms of bras across the different ages. And we realized that we are not going to be able to build one bra that everyone likes. So we designed it in a way where you can basically connect it to any bra of her preference. A wireless bra, a tank top to sleep, a sports bra, a surgical bra. and that's why we call it the Bloomer TAG, which TAG is the acronym for Tech Augmented Garment, is that we can augment the garment of her preference. Yeah, very cool. I'm quite intrigued by your go-to-market strategy.

21:07It sounds like you are not going direct to consumer and this is more of a clinical grade product. Did you consider going direct to consumer? And if so, like tell Tell us more about kind of how you made decisions around what market to focus on. We want to improve her outcomes. And the way for us to get there is to ensure that we are clinically validated along the way. And eventually we will be able to open more health and wellness channels. Some of the cardiologists that we've been talking to that are experts in AI, they've been saying we have the models now. We just don't have the data, right?

21:47So that's kind of the piece that it's missing. So we were sure that we wanted to bring good resolution, high quality data into what we're building and controlling the quality, making a regulated device, talking about your trifecta, AI sensors, and women's health. We added the regulatory side to it. If any of our listeners wanted to get involved in any of these studies or even potentially partner with you, what would that look like and how should they get involved? Yeah, so the easiest way to connect with us is through our website at bloomertech.com. There we have a form if you want to use our devices in your clinical studies or you want to be a subject in one of our studies.

22:35So whether you have already been diagnosed of heart disease or have family history of heart disease and you want to participate in those studies, that's a good place to sign up. Amazing. Okay, so we talked about the sensors, the data. Let's talk about AI. How are you incorporating machine learning in the bra and what does that look like? Give us some examples. Yes. So, well, in the early days, one of the things that we realized very early on, especially with ECG, is that for a long time we've been using unsupervised learning for a lot of things. Yeah, basically there's two main approaches in machine learning.

23:16One where you have supervised and basically have a ground truth labeled data set that you use for training and validation. And the other approach is unsupervised where you don't have the labeled data, right? Yeah. And you allow the data to gravitate towards each other to make definitions around it. Right. So in digital signal processing techniques, right, like when you use things like a blind source separation, where it's like if you think about right now with the microphone, there's one source of my voice. But if I have three microphones with all of that data, I can separate what is noise from what is actually my voice very easily.

23:57So for us in the way that we decided to use AI in our devices, it's a combination of things, right? So there's a thousand AI algorithms cleared by the FDA. Out of those, the majority are in radiology. Number two is cardiology. So what are examples of these cardiology-related AI algorithms that are FDA cleared? So examples include arrhythmia detection, right? So to detect tachycardia, bradycardia, AFib. The newer ones, I would say, have to do with aortic stenosis and a left ventricular ejection fraction, which is that the heart doesn't pump enough blood to the body. So it's an early indicator that you might develop heart failure and you can detect that from ECG data.

24:50Now you can detect sex from ECG data, right? Like just with the ECG, something that the eye wouldn't recognize. It has 97 % AUC. Which is area under the curve accuracy, basically score. Yeah. Yeah. So it's very accurate at detecting whether this ECG is from a male or a female. Yeah. Is the idea to create like real-time interventions? So you're wearing this bra and say you're about to have a heart attack. Does it detect this in real time? And then, I don't know, does it give you, the user, an alert? Does it alert a doctor? Like, tell us where the data goes and how is it being used? Well, that use case, I see it more as when in the future we have machine-to-machine communication that the device will be able to talk to another device to send, if she's going to have like sudden cardiac arrests to send someone to do CPR or something similar, right?

25:43Like I see it more on that spectrum. But the use case for our device, right? Like we start with cardiac monitoring, which is more traditional so that we can use the existing pathways. But we envision a world where for women that are at risk or diagnosed with heart disease, she can have this in her closet. If she's having symptoms, her insurance would cover her from wearing it. while she gets the appointment with her cardiologist. So especially for women with chronic conditions and women that want to prevent a recurrent episode that she already had a procedure, right? Like you can have this in your closet, you can have it accessible and you can collect very valuable data to prevent a secondary event.

26:27Amazing. So for now, it's a data collection device to kind of enable your doctor to see what happens between doctor's visits and hopefully be able to provide better intervention. For now, but the big picture is generating all those new digital biomarkers that are specific to women because there are heart diseases that are predominant in women that we've known for a long time. And we believe that most of the devices that exist today have been designed around signal and pattern recognition around traditional diseases that affect men more. So that's why women are having worse outcomes. Like now it's a good time to develop this next generation of ways to detect these patterns for these diseases that have been under-recognized for such a long time.

27:16After a short break, Alicia and I zoom out and talk about the potential of AI and personalized medicine as a whole. We'll be right back. Meet Nicole Nicholas, Capital One business customer and co-owner of Ansett Uncles, a plant-based restaurant and community space in Brooklyn, New York, that got its start from a need for unity. The inspiration, it was born from the desire to create a space that felt like home, where we can connect community culture, good food, and come together with family and friends. That's how we birthed aunts and uncles. Nicole and her husband, Mike, were fulfilling their dream of bringing people together out of their home kitchen, but they soon learned that the demand for community was greater than they knew.

27:59It became overwhelming and we were like, we need home, but not in our actual home. We realized that there was also a need in our community for something bigger and in our neighborhood. So we had to find a place. Moving from a home operation into a storefront was a huge next step, but Nicole and Mike were able to take it on with the help of Capital One Business. It's not for the weak. As a small business, finding resources is super important because that's the way you'll be able to manage and scale. We would have never done that without having Capital One to be able to help us along the way. The cashback rewards are very helpful.

28:35You know, it just gave us that runway to be able to breathe a little bit. Then you get to focus on the cooking of the food and making the experience great. To learn more, go to CapitalOne.com slash business cards. I just love the idea of wearables that empower people to really be the CEOs of their own health. I wear a lot of things and it helps me like track my sleep and my physical fitness. And, you know, at some point I wore a continuous glucose monitor. So I kind of imagine a future version of the world where I have an AI co-pilot for my health. I don't think we're there yet, but it sounds like that's kind of part of your vision as well.