“AI In Exercise Optimization, & Why It Matters”: Shikha Tandon with svexa) (Video + Transcript)

April 11, 2024

Former Olympian swimmer Shikha Tandon (svexa Chief Resilience & Partnerships Officer) delves into AI in Exercise Optimization; exploring its significance and implications in the field of sport, fitness, and well-being. Understanding this is crucial not only for fitness enthusiasts but also for healthcare professionals, researchers, and society as a whole.


In this ELEVATE session, Svexa Chief Resilience and Partnerships Officer Shikha Tandon discusses the role of AI in exercise optimization and its importance for health and wellness. The current landscape of human performance is heavily influenced by data, but there is a need for dynamic approaches to performance and health analytics. 

Like what you see here? Our mission-aligned Girl Geek X partners are hiring!

Shikha Tandon ELEVATE leveraging training optimization models allows you to adapt your training plans

Transcript of ELEVATE Session:

Shikha Tandon:

Thank you, Amanda. Hello, everyone. My name is Shikha Tandon. As Amanda said, I am the Chief Resilience and Partnerships Officer at Svexa. I’ll share a little bit about AI in exercise optimization and just discuss briefly why it all matters.

I plan to keep this high level, but more than happy to deep dive with anyone one-on-one after. Send your questions in. Happy to get to them at the end. If not, happy to chat individually. I’ll preface all of this with the fact that exercise and activity has a direct impact on a person’s health and wellness.

Before I begin, I was asked to provide some snippets of my personal journey thus far and how it has brought me to my current role. I was also told that most speakers don’t spend enough time as they should on this slide. I’m going to spend a few minutes here.

I grew up in India. I spent most of my school and college days as a student athlete. I represented India for almost 15 years in swimming, including at the Olympics in 2004 in Athens. And through my journey as an athlete, I was always fascinated with the bioscience.

As an athlete, you’re constantly trying to push the limits of what’s possible. You are introspecting a lot, and so that’s what I ended up choosing as my majors in college. As an athlete, I had also been drug tested a lot. I knew that… That was something that I wanted to work in in anti-doping science. The only problem there was that there was absolutely no such career option in India, and that prompted me to move to the US for a second masters.

As you can imagine, anti-doping science is not a course available in college. During this time, I had to customize my projects to gain specific experience in this field that would set me up for hopefully a role after. After I graduated, I was fortunate enough to get my dream job and work at the USADA, which is a United States Anti-Doping Agency as their science program lead.

It was a research and education focused role, but I also had the opportunity to dabble a little bit in product management as I built one of their first online education modules for health and medical professionals that were working with athletes.

Fast-forward a few years, I moved to the Bay Area, suddenly surrounded by everything tech and something that I had no immediate interest in at the time, but it grew on me. As one does, I attended multiple conferences like this, some in-person, trying to figure out what next.

During one of those conferences, a presentation on human performance and analytics got my interest. I mentioned to the speaker after that if he did anything with that research that he presented, I would love to be involved.

Leveraging my experience at USADA and a little bit of the PM work that I did, I worked in product for a few years as a product manager at different fitness wearable companies, and that was a time the wearable industry was picking up. I had a great chance to just learn a lot in a short span of time.

After that, I went on to work as a product manager for TechCrunch, a tech media company where I learned a lot about the startup ecosystem. By now, it had been a few years since that conference that I just mentioned. Turns out the speaker co-founded a company, Svexa.

For the past few years, I’ve been at Svexa now… was one of their very, very first few employees moving up and across different roles within business and partnerships.

My current role, the way I see it, is the resilience aspect is just a really nice mix of learnings from my time as an athlete, but also while driving the company business and the partnerships forward. I know this session was listed on the tech track and that there is a separate one on career, but if anyone wants to learn more about how and why I made all these seemingly unrelated career moves across titles and roles, please feel free to reach out to me and happy to go deeper.

What is Svexa? Svexa, it stands for Silicon Valley Exercise Analytics. It’s an exercise intelligence and human performance company. What we do is we develop proprietary algorithms that we then license out to support health and fitness companies.

Think of us as an ingredient brand similar to Intel Insights, but for human performance optimization. We are a B2B company. Our algorithms are applicable to athletes, sports teams, irrespective of level, sports tech, wearable tech companies, corporate health, virtual health, hardware, or software. Just essentially anyone that has health and wellness data on their end user and is hoping to offer some sort of individualized analytics and hyper-personalized insights from this data.

The current landscape when it comes to human performance is already pretty heavily influenced by data, but the key, really, is in terms of understanding this, is that this landscape is continuously evolving and that really requires a dynamic approach to performance and health analytics. Bit of background, over the last 10, 15 years, there’s been a huge tech revolution in terms of what’s out there.

It is really never been easier to gather data, whether you’re an athlete or just someone looking to stay active and healthy. This has led to an overwhelming amount of data with very, very limited insights and recommendations that are personalized to the specific user in their context. Today, most technology displays aggregated data or just provides insights and recommendations that are based on either general population or just limited data streams. And this is not scalable or even flexible across industries.

When we look forward in terms of what is this need, we’ve seen that there is a trickle-down effect from elite sport across industries. There is this need to deliver actionable hyper-personalized insights that are scalable across industries. As an example, 15, 20 years ago when I was competing and training, heart rate monitoring was accessible only to elite athletes. And even within our team, we had one or two heart rate monitors that we would share among everyone on the team within that session.

Today, almost every smart device has this capability and it’s almost expected as a minimum feature for users whether or not they know what to do with it.

Diving a little deeper into the tech landscape, as I mentioned in the previous slide that wearables and heart rate devices are a great way to gather data, but primarily, or most of them, are single or limited source. These could be smartwatches, CGM trackers, smart rings, and many of these have accompanying software or hardware which then enables data visualization for their B2C clients and end users.

If we go a step further, so you have platforms such as athlete management systems that can import data from multiple sources, but typically, they still all fall under a data visualization tool. They may offer some sort of trend analysis or population-based comparison analysis for their end users.

When we move towards really individualized insights, this typically falls on some level of human intervention and expertise. And in the sports world, this may be a professional coach. For fitness apps, this may be a coach on the platform. This, however, is not scalable and there’s only a certain number of athletes or users, a coach or trainer, can manage.

While many of the existing tech could be used by these professionals, personalization is, on some level, limited to the knowledge of the individual to interpret the data. This may or may not be relevant across industries. For example, the coach may be in a position to deep dive, may not be in a position to deep dive into the health metrics, while the health professional may or may not feel comfortable with training optimization.

Today, we have… AI solutions are leveraged to counter some of these issues, but just adding AI to the mix is really not the answer. That’s the part that Svexa is building. The technology is a combination of both AI and human domain expertise. The AI aspect is utilized to scale the algorithms and the offering.

If you think about from the competitive lens, we have hundreds of algorithms that we can license out and different combinations of these could be licensed to add value to most of these existing solutions, whether they are hardware or software. These algorithms then can be used to either make all of these offerings more scalable or just applicable across industries or providing deeper insights or just looking into them at a very, very hyper-personalized lens.

When it comes to exercise optimization, multiple factors such as sleep, nutrition, mood, stress, activity, travel, injury, illness, and any other existing health conditions, they all play a role in an individual’s readiness and ability to perform and be productive on any given day. And today, we have access to so many different technologies. These inputs could really come from various different sources. A lot of them are reported differently. Some could be subjective in nature; some could be objective in nature in terms of the data.

The key is to be able to handle an account for all of these, keeping in mind the individual at the heart of all of this. And in some instances, similar metrics are gathered by multiple devices for the same individuals. For some of us, we may have multiple apps on our phone that are giving us daily readiness scores, for example. How does a score of an 85 on one app compare to a score of a 75 on another? And what does one even do with this data?

As we think about addressing this need to deliver the scalable intelligent solutions, at Svexa, for just ease of licensing, we refer to groups of our algorithms as products. One of these products is called the Athlete Passport, which is essentially a concise representation of all the data available for an individual. Unlike existing tech, this is not just a visualization tool, it goes well beyond it. It highlights key response patterns and correlations between metrics for that specific individual. For example, it could be correlations between mood and stress, or sleep and nutrition. This enables output and insights that are more actionable.

For example, some people have great sleep after a hard workout, while others may have restless sleep if they’re very tired. If that person has travel somewhere, so add travel and jet lag to the mix and then it impacts these metrics. If you add suboptimal nutrition and water intake due to the said travel, the equations change again. Instead of just stating that sleep quality is trending a certain way, we can explain specifically for that individual what is driving these changes and how and what to do with it.

Again, building further on this Athlete Passport, we have a Digital Twin technology as well. While Digital Twins are gaining traction across biomechanics, you also have some for in-game and on the field strategy, and also maybe some in the medical field. Ours focuses on the overall physiology of the individual. These Digital Twin algorithms enable us to simulate millions of possible scenarios to generate optimal recommendations for training performance and health.

Some examples of how powerful these optimization models and recommendations engines are… Just last month, the technology predicted a half-marathon time for one of our team’s data analyst with 0.2% accuracy and he just became the second-fastest European ever in this event. Pretty exciting. For the past few years, we’ve had similar accuracy across sports, not just with elite athletes, but also with recreational athletes. And not just with event timings, but also for personalized heart rate zones and things like that.

We’ve talked a bit about exercise optimization and how we go about it, but why in all of this, how does it even matter outside of elite sport? We are functioning as an intelligence layer between the data layer and the interactive layer. We can tell anyone what the optimal amount of activity is for them at any given point of time.

This intelligence is really the product and it’s contextually driven by a combination of algorithms. Think of Intel, it’s a seal of excellence for laptops. Similarly, you have GORE-TEX, which then functions as an ingredient brand for others like Patagonia, North Face, Arc’teryx, which are all similar products, but GORE-TEX is able to service them all and doesn’t really play any direct part in their final design, but essentially powers all of these different brands.

Svexa tech is data source and device-agnostic, and we can work with as little or as much data that is available. This is licensable to both, as I mentioned, hardware and software. Essentially, we’re democratizing access to health and wellness optimization through our team’s deep understanding of human physiology. You have generative AI and voiceover tech, it’s getting a lot of traction within the health and fitness space. Fom our perspective, all of these fall under the custom front-end implementation.

Think of all of these talking to that Svexa intelligence model before they provide any sort of output. And of course at a much, much broader level, the United Nations Sustainable Development Goal 3 aims to ensure good health and wellbeing for all at all ages.

Individualized recommendations for physical activity and exercise are pretty powerful tools when we are trying to achieve this milestone, and so we are striving to play a small part in this.

Going beyond, when you think of elite athletes, at one end of the spectrum, you have learnings from this can be extrapolated to recreational athletes, corporate health, and general health. Whether you’re recreational athlete training for your first 5K or running a marathon, leveraging some of the training optimization models allows you to adapt your training plans in a way that you can maybe prevent over-training or prevent injury fitness enthusiasts.

We’ve heard the generic recommendations of 10,000 daily steps and 150 minutes of activity a week. While all of these have been great in terms of getting people active and moving, going forward, I think adaptation within this that factor for everything else going on in your life, all of that has been shown to be more effective than just following a one size fits all model.

Even if it means helping you pick the next workout on one of your favorite fitness apps. When we look at corporate health, if we take a step back and replace performance for productivity, we could potentially utilize the same algorithms that we use for injury and illness prediction to positively make an impact on burnout and productivity within corporate health.

The next piece of the pie is health and wellness, and exercise and activity plays a pretty crucial role in management. The same algorithms that are used to assess an athlete’s daily readiness, for example, via the multiple input parameters that I mentioned, they can and are being actually contextually applied-

Amanda Beaty:

All right. Sorry, Shikha, we’ve got to head over to the next session.

Shikha Tandon:

Okay. Thank you so much.

Amanda Beaty:

If you have time, when I close this, you can pop over to the Q&A and look what’s in there. And the attendees won’t be able to see it, but you can, if you want to put the relevant answers on LinkedIn or something like that then…

Shikha Tandon:

Okay, sounds good. Thank you.

Amanda Beaty:

All right, thanks everyone. We’ll see you in the next session.

Like what you see here? Our mission-aligned Girl Geek X partners are hiring!

Share this