Evan: Welcome to the Mining Your Business podcast. I’m your host, Evan Wimpey, and I’m excited today to introduce Olga Sazonova. Olga is the director of data science and analytics at NutriSense. There, she leads a team of analysts and engineers and they focus on NutriSense’s promise, which is to deliver personalized nutrition and wellness and disease prevention. Olga’s background in biological data science. She’s previously worked at GRAIL and 23andMe. She’s got a Ph.D. in biomedical engineering from Boston University, and we’re excited to have her on the show. Olga, welcome to the podcast.
Olga: Thank you so much, Evan, and thank you for that wonderful introduction.
Evan: Sure. Yeah. And that was touching the surface. But maybe can you give us a little bit of background on where you got from your biological background to where you are now and data and analytics and the current role that you have at NutriSense?
Olga: Yeah, sure. I’d be happy to. I started my interest in science very young when I was even away back to high school and the science fair program. That got me excited about wanting to understand how biological systems work. And I pursued a Ph.D. in biomedical engineering, but most of that work was actually on the bench. And I didn’t start becoming what is now called a data scientist until the end of my Ph.D. when I wanted more sophisticated ways to analyze the data that I was collecting. And that convinced me that these computational approaches are very useful in science. And that’s why I pursued a postdoctoral fellowship in computational genomics. So that was kind of the early evolution. And at some point, I realized I didn’t want to be in academia. I entered the industry as a computational biologist working mostly with genomic data. I spent several years at 23 and me working on risk prediction and how we can use a person’s DNA to assess whether they are at higher risk of developing a particular disease. I worked at Grail looking at blood-based cancer detection, also using both DNA signals floating in the blood and proteins that are floating in the blood that may be derived from a tumor. And that, you know, one thing that I loved at 23 and me was working for a direct consumer product, and that was something that I missed once I left. I liked taking complicated biological signals and turning them into something meaningful that a consumer can derive value from and be delighted by. When I was looking for the next chapter in my career, that was really what I was looking for. I was excited to be at a smaller startup. I was excited to be in a role that would touch many aspects of the business rather than focusing on something that’s very much a research project. And for me, you know, the CGM space was up and coming. This idea is that we can take a tool that was previously for sick people, you know, acceptable to use if you’re a diabetic. And now we want to deliver the value of the same tool to everyone. That story appealed to me. And so it was kind of a perfect synergy that I found myself at NutriSense where I get to think about interesting biological questions. I got to think about making a direct-to-consumer product based on biomedical data. And of course, there’s the business, intelligence, and analytics to help us grow as a company that is relatively new for me as a domain. And I’m getting a lot of knowledge in this role.
Evan: Awesome. Yeah, that’s very exciting. And it sounds like a place in a role that is very well suited to your background. I am curious. I’m not sure how old NutriSense. You mentioned start-up. It’s been around for at least a little while now. My impression is that – Go ahead.
Olga: No, I was going to say the company started just a few months before COVID. So they’re pretty young.
Evan: Perfect timing.
Evan: All right. And I take it they generate a lot of, they collect a lot of data and the sort of purport to help give that real personalized wellness information. So they’re no strangers to analytics. They’re sort of almost built for analytics. Do you find that to be the case? Do they feel analytically ready, or maybe more data appreciative than previous places that you’ve been?
Olga: Hmm. I think there are a couple of questions built into that, so let me unpack it. What we offer our members is their data. We offer a way to detect something that is otherwise very difficult and frankly, painful. I don’t know if you’ve pricked your finger much, but it’s unfun, difficult, and painful to track. And we offer, in addition to that dietician support that can help people get the most out of that data. So it’s very much we are a data-driven company because the product we’re offering the value is data for you to understand yourself and better understand your metabolism. But we kind of add abstraction and benefit from the interpretation that anyone can work with a dietitian to get. So yes, by that being our core product, we are a very data-driven company. We, of course, from day one have been collecting data from the CGM that people are using. But we are also very data-driven in our marketing and our operations. Because members work, most of our members work with registered dietitians. Dieticians need some information about these people’s well-being to do a good job of guiding them. So in addition to the food blogs that our members often keep by working with the registered dietitian, we will end up learning more and more about someone’s health through the data they volunteer to get the most out of that relationship. So we have questionnaires on people’s health. I think this is a much richer subject, but the interactions between members and dietitians can reveal many, many interesting insights about a person’s lifestyle habits, their, you know, their well-being overall, how they respond to various modifications of their lifestyle, and so forth.
Evan: Yeah. That’s very exciting. From the little I know about nutrition, it’s, it’s CGM, it’s continuous glucose monitoring. But yeah, certainly as a nutritionist, that’s not the only piece of data that you want. So maybe we can just talk a little bit more about the types of data that are offered. And when you mentioned the conversations or the interactions that they have with nutritionists, I think is there an opportunity, is there text or speech there that you can process or try to analyze?
Olga: Absolutely. So from a data scientist perspective, this is very exciting to me because all of the interactions are through text and they are captured in our app. And so we are developing probably the richest corpus in the world of interactions between dietitians and the general population. And you can segment that population however you want. We work with individuals who have a Type two diabetes diagnosis. We also work with pre-diabetic people. We work with many people who have no indications. And on the other end of that spectrum, we work with people who are very, very, very fit and who are using our product to optimize their fitness to levels beyond which I could ever aspire to. So the fact that all of this happens, not through a zoom call but by text, is a really incredible opportunity to learn, you know, how these interactions unfold, what value a dietitian can bring, how the complexity of the conversation impacts the outcomes that the member experiences. That’s a hypothesis. That’s something we haven’t tested but are very interested in. So there’s I think we’re just starting to scratch the surface of what we can do with that data.
Evan: And that’s exciting. That’s certainly a lot of data that you have on folks. And I want to ask you about so I think on your team, you’ve got both analysts and you have engineers as well. Can you talk maybe a little bit about the challenge of having all of this data in particular that the, I don’t know if it’s streaming data or if it’s processed in batch, but you’ve got a lot of data that comes in just from the CGM on, folks? Are there challenges associated with that?
Olga: Well, I think we benefit from the fact that even though our dataset is impressive, I would say that Netflix and Amazon have had to solve much bigger data collection problems than we have, and so best practices exist for a lot of what we need to do. You know, certainly building these systems from the ground is non-trivial, but I don’t think that our CGM technology creates any unique difficulties for the kind of upstream all of the analytics work to collect the data to pipe it in. Now we have a very broad range of data sources and I think that is certainly keeping us busy in terms of having integrations for all the different kinds of things we measure. I mean, you know, we measure all the standard marketing analytics on ad spend, clicks and we measure how people interact with our website. And at the same time, we measure every event that happens in our app where a user opens the app and they tap here and they tap there and how long they see any given feature. So I would say that the breadth of that is perhaps as challenging to wrangle as the amount of data. Now, you know, ask me in a year when we have grown our user base by many folds and I may tell you a different story. And of course, as I said before, you know, we’re pretty young in terms of the age of the company. So the challenges will change. I’m certain of that. Yeah.
Evan: Certainly. I’m curious about this so you mentioned a couple of times you’ve got sort of your health and wellness related data and then you have your traditional data that any business faces sales, marketing, financial data. And your team oversees analytics for both of those. Is there a clear line, or are these the analysts that have a biological background and they focus on health and wellness and these folks are sales and marketing types folks?
Olga: We’re not that differentiated at this age know the stage of the company and the team. We need everyone to wear hats. My background is obviously very biological and biomedical. I’m most comfortable in our research space. So, I’ve been looking to build a team with skill sets that complement where I can guide how to design the research study, how to adequately test certain hypotheses, and the statistical methods we’d want to use. And I get to partner with people who can provide me guidance on the best way to transform data in and from an unstructured source so that it can sit in this relational database and be accessible through these downstream tools and so forth. So, you know, are we differentiated in terms of how this analyst works on this type of problem and that type of problem? No. But do we bring complementary skill sets? Yes. And we need to execute on the very broad range of problems that we need to solve.
Evan: Yeah, I think. I think it’s perfect. There’s not just the data scientist who knows the data science skills, the rigor, and the background certainly is important. Yeah. So a common challenge that we talk about on the show and it’s not – you’re the first guest we’ve had to talk about it in health and wellness data but is trying to get some insight, some analytics to drive real change from somebody. So like in the retail space or the manufacturing space, it’s changing the way they do business. I’m curious if the analogy for your role, you’ve got a nutritionist who’s offering health and wellness advice to folks. You’re analyzing a lot of the same data that they have and they see. Is a nutritionist sort of your end user that you’re trying to help influence or help give them insights that the data is telling?
Olga: No, you know, the dietitians, the registered dietitian team is part of the service that we offer to our members. But our fundamental focus is on everyday consumers who are interested in learning more about their metabolism, or they have some concerns because metabolic disease runs in their family or they’ve been diagnosed and they want to make a change. And we are offering them the combination of real-time insight into their metabolism and an expert to help them make the most of that data in order to help them achieve change in their life. So, yeah, we are also chasing change. We’re chasing behavior change. And from my perspective, you know, working at 23 and me, the goal is very similar. The hope was that if we teach people about their genetic risk and tell them what they can do to mitigate that risk, they will be motivated by that information to do it. And the truth is that some people are, but a lot of people are not. You know, it’s very abstract. If I tell you that in 30 years, your risk of getting X disease is 50% higher than mine, but the actual risk across the entire population is 2%. What do you do with that? But if I tell you, Hey, you just ate a cupcake, and here’s what happened to your blood sugar. The next time you eat a cupcake, eat some cheese first because we know that a combination of protein and fat can help you blunt that response. I’ve just given you very personalized insights about yourself that mattered today, not in the future. And I’ve given you something very actionable to follow it up on. I think that’s a very powerful tool for behavior change. And together with a dietitian who can give you a much more, I’d say, nuanced explanation of what I just said, that’s I think, the winning combination of resources that we want to give everybody who joins us for our programs.
Evan: Yeah, I love it. I think you hit the nail on the head. You want the output that you provide to be a decision-able, to be actionable by someone. Exactly. And so being clear and immediate is good. I’m not. Not. I’m not hoping this is the case, but certainly in other industries it’s been the case where the industry expert or the dietitian may have a different viewpoint than what the data analytics can come up with. Have you noticed or has there been a case where there’s been any conflict where the dietician says, “no, no, no, don’t, don’t do what the data spits out?”
Olga: Well, one of the strategies that Nutrition took very early on is to build a team of dietitians and nutrition professionals who are going to be trained on the technology we use, who is not going to necessarily be wedded to the traditional models and school of thought in dietetics, but are going to be very open to being data-driven. And our team even has a mantra of data over dogma. And I think so within our company, we’re very aligned on the idea that you want to practice a holistic approach to someone’s well-being. So the CGM alone is not the answer, but the CGM alone does give very uniquely useful insight. And if what you see in the data goes against what you’ve been taught in your degree program, follow the data. You understand what may cause those conflicts and don’t necessarily dismiss them. So a very personal example. I’ve been running these experiments on myself where I practice intermittent fasting and then I’ve been deliberately breaking my fast with dates. Now dates are very, very sweet, but they’re also traditionally thought of as a great breakfast food, and people with diabetes are even encouraged to consume dates in favor of other forms of sugar. So I was curious to see, okay, what will they do? And I learned that they indeed make my blood sugar spike as badly as if I had eaten that cupcake. So there’s one myth out the door, at least for me personally, that this is a good breakfast option. And then I had a lot of fun trying to modulate that glucose spike by eating different foods together with the dates or different sequences. And I was working with my dietician the whole time to understand like, you know, why didn’t dates with butter work? So good. I can’t even express to you Evan how much joy I got out of that experiment. It was like, the most delicious thing. But it didn’t help at all. Well, it could be because fat alone isn’t enough for Olga to blunt that sugar response. Maybe Olga needs more protein. And so we ran those experiments. The point I’m trying to make is that I found our dietitians to be very aware of the value of the data, but also kind of open to learning with the members as they go, as well as relying on the great body of literature that exists around these things.
Evan: Yeah. And that’s perfect. And that’s, I think, a partnership that most folks would strive for leading up a data and analytics team for the people that they work with. They’re sort of business partners there. They are accepting and want to work with the data folks.
Evan: So you mentioned experiments a couple of times. And, you know, a lot of things in data. You know, certainly, one data point we don’t want to draw a lot of inferences from – does that translate to glucose monitoring as well? Like if you’ve tried dates with butter and you got this result, is it reasonable to expect that results all of the time, or is more data better? And you can see there’s sort of probabilistic there.
Olga: So for me, if I ate dates with butter three times in a row, what would happen? Would I expect the same result? I think for every individual there are so many other factors that go into impacting the results of a given meal, that it would be wrong to expect the same results every time. For example, if I slept 6 hours versus 8 hours on a given night, I should see a different response to those dates. Female hormone cycles influence our metabolism significantly. So I think to get an accurate picture of what dates with butter, and how the dates of butter impact my metabolism, I should repeat the experiment under different conditions. I will tell you though honestly, that can be pretty tiring, you know, having to come up with a space for all the different experiments one might do. And this is where I think being a scientist, you know, I make my own life harder than it probably needs to be. And I could get a pretty good approximation without being so rigorous. But I think the other element of your question, like, does that mean that Evan eating butter with dates should expect no benefit from the butter? And the answer is no. And this is one kind of amazing application of continuous glucose monitor is that we are seeing evidence of unique postprandial, post-meal responses to the carbon sugar intake. And so what part of the value premise of our company and our product is like, you should learn. You should go out and take this tool and learn, you know. For you, is it brown rice or white rice that seems to be worse? We’re all told that brown rice is the best. But for me personally, I seem to spike more with brown rice, which is totally blowing my mind. But the data are undeniable. And yeah, I’ve repeated that experiment and it’s still consistent, so go figure.
Evan: Yeah, that’s really exciting. It’s good to know, and it’s good to know as a consumer. You know, if I were a customer, I would want to be really sure about the cookies and the ice cream and the dates with butter. So I’m going to repeat a lot of experiments with this, just to be sure.
Olga: Just in case. Yes, that’s right. I mean, but the same meal eaten four hours later in the day can also have a different impact because our insulin sensitivity changes just with circadian rhythms. So your body’s response to the same input will be different because you’re closer to your sleep cycle or not.
Evan: Yeah, very fair. And I think that’s probably intuitive for a lot of people and it’s good to get the data behind it to make the best decisions that we can. I want to ask you one last question. So you’ve got a pretty rich background in the sciences. You’ve been here for a little while now. If you could choose any type of, maybe, maybe not experiment, but any type of effort that you wanted to try to dig deeper and uncover something important that’s sort of not known or there’s not enough time or resources on it. Now, where would you like to, assuming that all the dieticians and everybody nutritionists are on board with your vision, you get to pursue whatever you’d like and point your team to pursue whatever you’d like. Is there a burning question that you’d go after?
Olga: You know, there is one question for me and it’s a question for, I think, other people in this space who are being really honest with themselves. It’s a question I had when I started this role when I was contemplating starting this role, and it’s still an open question. And the question is: for healthy people who don’t have any signs of metabolic dysregulation, is this actually going to benefit me in the long run? I may delight in learning more about myself, and I may change my behavior somewhat, like I’m not going to break my fast with dates, but am I actually better off? Like, is my lifespan going to extend? Am I going to reduce my risk of cardiovascular disease, metabolic disorders, etc.? And that’s, you know, that’s no secret that the field doesn’t know this, that if you go to your doctor and say, I am healthy, should I wear a CGM? They’ll say, I don’t know. I don’t. No one’s proven the benefit. So I think NutriSense and every other company in this space, that’s our biggest mission. If our goal is to help people improve their health and well-being, we need to prove that the tools we’re offering actually accomplish this. And I think we can all bring anecdotal evidence to support that claim. But until you’ve done the right experiments and the right trials, you know, the right studies, they’re just claims. So I think there’s a question of what’s good for the business and there’s a question of what’s good for society. They don’t always align. And I think once NutriSense is in a more mature place, I would love to tackle that question head-on. And then, of course, we can take any of the steps even further back and say, well, some of us are very healthy today, but we’re going to get less healthy in specific ways over time. And that’s where genetic prediction can help, too. So find the people who are most at risk of developing, you know, early onset diabetes, type two diabetes, give them a CGM, show that it’s a great behavioral change tool that teaches them how to prevent the onset of the disease they’re at risk for. And I think proving that that could work would just change the paradigm dramatically. And that gets me very excited.
Evan: What an exciting space to be in. What an exciting problem to be working on. Well, good. That’s all the time we have. I want to thank you so much for coming to the show. You’ve been very insightful.
Olga: Well, thank you, Evan. It’s been a pleasure to speak with you.
Evan: All right, folks, if you’re interested in working in this space, check out NutriSense. If you’re interested in helping generate more data, then be a NutriSense customer and get your glucose monitor. If you’re interested in this topic, then like and subscribe and catch the next episode of Mining Your Business. Olga, thanks so much.