AI transcript
0:00:02 (upbeat music)
0:00:05 (crowd cheering)
0:00:09 – Pushkin.
0:00:19 What’s one surprising thing your work has taught you
0:00:22 about elite athletes?
0:00:24 – I never thought I would see muscles
0:00:26 that were so developed.
0:00:28 (laughing)
0:00:30 They broke our scale.
0:00:30 – Wow.
0:00:32 Yeah.
0:00:33 – Like it was just too big.
0:00:35 The machine, the AI couldn’t figure out what it is.
0:00:37 – Well, no, the AI found it,
0:00:40 but we’re like our kind of rating system.
0:00:40 – Wow.
0:00:43 Was there a particular athlete or a particular sport
0:00:44 or a particular muscle?
0:00:46 What muscle broke the scale?
0:00:50 – The gluteus maximus breaks it at fair amount.
0:00:51 – No kidding.
0:00:52 – Yeah, yeah.
0:00:53 – Fantastic.
0:00:53 – Yes.
0:00:54 It’s a pain in my butt.
0:00:56 (laughing)
0:00:58 Like ’cause it’s too big?
0:01:01 – Yeah, yeah, it’s just so big.
0:01:03 (upbeat music)
0:01:09 – I’m Jacob Goldstein and this is What’s Your Problem.
0:01:11 This month, a bunch of Pushkin podcasts
0:01:15 are coming out with Olympics-inspired shows.
0:01:18 Revisionist History has a series about America’s decision
0:01:23 to participate in Hitler’s Berlin Olympics in 1936.
0:01:25 The Happiness Lab has an interview with a coach
0:01:27 who coaches coaches.
0:01:29 And here on What’s Your Problem,
0:01:31 we’re gonna be talking with people
0:01:34 who are working at the frontiers of technology
0:01:37 to help elite athletes perform better.
0:01:42 For example, today my guest is Sylvia Blemkert.
0:01:45 She’s a professor of biomechanical engineering
0:01:47 at the University of Virginia.
0:01:48 And she’s the co-founder of a company
0:01:51 called Springbok Analytics.
0:01:53 Sylvia’s problem is this.
0:01:56 How do you combine MRI scans and artificial intelligence
0:02:00 to generate new insights that can help both elite athletes
0:02:04 and people suffering from diseases that affect the muscles?
0:02:07 Springbok’s clients include medical researchers,
0:02:09 Olympic athletes, major league baseball,
0:02:13 and several professional basketball and soccer teams.
0:02:15 You’ll hear about all that on the show.
0:02:17 But first, we’re gonna pick up
0:02:19 where we left off in the conversation.
0:02:23 We were discussing the extraordinarily large muscles
0:02:25 of elite athletes.
0:02:29 And then Sylvia told me something even more surprising.
0:02:32 – The other thing is that they have some tiny muscles too.
0:02:36 – Huh, like smaller than a normal person’s muscle.
0:02:38 – Yes, much smaller.
0:02:40 They put their muscle where they need it.
0:02:41 – What’s an example?
0:02:44 Like what muscle is tiny and what kind of athlete?
0:02:48 – Calf muscles are small in most fast athletes.
0:02:54 Like you look at a sprinter or like a running back.
0:02:56 – It’s just all quad, no calf?
0:02:59 – All like thigh, no calf.
0:03:00 Yeah, thigh and hip.
0:03:03 It kind of makes sense because if you’re trying to run fast,
0:03:04 you wouldn’t want to put a lot of mass
0:03:06 like at the end of your leg.
0:03:09 It’s like adds a lot of inertia to like move your leg.
0:03:12 Because the muscles are important for sprinting.
0:03:15 That’s the interesting thing, but they just don’t.
0:03:16 They’re small, they’re very small.
0:03:18 – Uh-huh, uh-huh.
0:03:20 So I’m particularly interested at this moment
0:03:24 in the sports piece of what you do.
0:03:25 – Uh-huh, good.
0:03:28 – I’m curious, by the way,
0:03:33 do you work with any Olympic teams or Olympic athletes?
0:03:35 – Yeah, yeah, we’ve actually been working
0:03:38 with several different Olympic athletes.
0:03:41 The ones that probably that come to mind most
0:03:46 are multiple players on the U.S. women’s national soccer team.
0:03:46 – Oh, cool.
0:03:50 Tell me like, tell me the story of that work.
0:03:53 So they came to you, what did they,
0:03:54 what did they want when they came to you?
0:03:55 Like how did that begin?
0:04:00 – They came to us along with their team.
0:04:02 So the technology we provide, you know,
0:04:04 an athlete could understand it,
0:04:07 but really with their team to help them figure out
0:04:10 how to keep athletes healthy.
0:04:12 – So what did they say?
0:04:14 What did they say when they came to you?
0:04:19 – So, for example, one athlete that’s coming to mind
0:04:23 had a known imbalance, side to side,
0:04:26 that based on a history of injury.
0:04:28 And they really wanted to know
0:04:29 where that imbalance was coming from.
0:04:32 – So the woman had hurt one of her legs
0:04:35 and that leg was, even after she came back,
0:04:38 that leg was weaker essentially than the other.
0:04:40 I mean, is that the sort of gross, you know, macro view?
0:04:42 – Yeah, that’s a fair way to say, yeah, exactly.
0:04:44 That’s a nice way to put it, yeah.
0:04:46 And they wanted to sort of find her like,
0:04:48 okay, but we can see that,
0:04:50 but what’s going on on the inside?
0:04:52 Like muscle by muscle, tell us that.
0:04:53 – Yes, exactly.
0:04:56 That’s precisely what we do, we go on the inside.
0:04:58 ‘Cause on the outside,
0:05:03 you see perhaps that her knee extensor or quads
0:05:04 seem weaker on one side than the other,
0:05:09 but there’s four quads, quadriceps, four muscles.
0:05:11 And so it’s not clear which of those muscles
0:05:14 are actually the culprit for that imbalance
0:05:15 and in what way.
0:05:17 – Good, so this is their question
0:05:20 and then what happens next?
0:05:23 – So the first step is an MRI scan.
0:05:28 And so with these athletes or teams,
0:05:33 we have ways to connect them with an MRI machine,
0:05:36 whether it be through an imaging center
0:05:37 that they partner with
0:05:42 or we’ve even actually brought MRI mobile trucks to sites
0:05:44 to make it easier for them.
0:05:46 – It’s like the players run off the field
0:05:49 and get an MRI and go back and keep playing, yeah.
0:05:51 – Yeah, kind of, yeah.
0:05:53 It helps just with the timing of things.
0:05:56 But so first we connect them there.
0:05:58 So it takes about 10 minutes.
0:06:01 Then they send those pictures up into the cloud
0:06:04 into our server.
0:06:06 And then we crunch through it
0:06:10 and then we send back a report on their muscles.
0:06:13 We also have what we call interactive viewer
0:06:17 and it’s presented in the form of a 3D model,
0:06:18 three-dimensional model.
0:06:21 So you actually see your own legs,
0:06:23 the muscles and bones, your own muscles and bones
0:06:27 that we’ve identified from the images,
0:06:29 going through a process called segmentation
0:06:32 where we find all the muscles and bones
0:06:33 and then we reconstruct them.
0:06:36 So it’s kind of like a digital twin of that person
0:06:38 that they can see on their computer.
0:06:41 And so along with it are all these metrics
0:06:44 that helps them understand their balance,
0:06:48 the development or strength of the muscles
0:06:50 and the health of the muscles.
0:06:52 – So tell me about this report they get.
0:06:54 Like what does it say?
0:06:58 – So the basis of that is actually a lot of research
0:06:59 that we did over many years
0:07:03 because you need to understand
0:07:05 where somebody falls relative to a normal,
0:07:07 essentially to give them,
0:07:10 essentially we have a scoring system for the muscles.
0:07:12 And that’s based on comparing
0:07:16 with a large data set of healthy individuals.
0:07:19 And so we know for a given person
0:07:24 based on their sex, age, height and weight,
0:07:27 how big we expect all the muscles to be.
0:07:29 And that’s through a lot of previous research.
0:07:32 So then we can say, okay, here’s where you land
0:07:36 each particular muscle compared to this,
0:07:38 what we call a normative database.
0:07:40 So we call it a spring box score.
0:07:43 – Do you do it for every muscle in the leg or?
0:07:44 – We do it.
0:07:46 So our primary product that we started with
0:07:51 was every muscle in the legs, essentially from belly to feet.
0:07:53 No muscle left behind.
0:07:54 They’re all important.
0:07:56 – How many muscles are there?
0:07:58 – 35 per leg.
0:08:00 So 70. – Okay, okay.
0:08:01 More than I would have guessed.
0:08:02 But what do I know?
0:08:03 – There’s a lot in there.
0:08:04 – Yeah, there’s a lot in there.
0:08:06 – They’re all important.
0:08:11 – Okay, so it’s basically how strong and healthy
0:08:16 is every one of those 70 muscles relative to baseline?
0:08:19 – And then the asymmetry comes
0:08:20 where you can compare side to side.
0:08:22 So for each of the 35 muscles
0:08:24 that they exist on each leg,
0:08:26 we can say which side is bigger,
0:08:28 which side is smaller, and by how much.
0:08:32 And then we also have normative values for that
0:08:34 because we’re all just slightly asymmetric, right?
0:08:37 – And presumably some muscles are more asymmetric
0:08:38 than others.
0:08:42 And so you wanna know kind of how asymmetric relative
0:08:45 to baseline is this particular pair of muscles.
0:08:46 – Exactly, yeah.
0:08:52 – And so in the case of this soccer player
0:08:56 who came to you who knew she had some kind of problem
0:08:57 with her quadriceps on one side
0:09:00 but didn’t know what was going on, what did you find?
0:09:03 – We found some imbalances,
0:09:06 actually not just in those muscles.
0:09:09 It turns out that it’s all connected.
0:09:14 So if you have a weakness or an imbalance
0:09:16 in one set of muscles, usually some other set of muscles
0:09:18 are compensating in some way.
0:09:21 – Well, it’s like when you mess up,
0:09:24 even if you’re just a recreational athlete, right?
0:09:25 Like if you mess up something,
0:09:27 you mess up your ankle, then you start walking funny
0:09:29 and then like your back hurts
0:09:30 ’cause you’re walking funny, right?
0:09:33 Like that is a very anecdotally apparent thing.
0:09:34 – Yeah, yeah, we all know that,
0:09:36 but it shows through in the scan.
0:09:39 But the thing is that it’s not very intuitive
0:09:43 from the outside which muscles have been affected
0:09:44 and how they’ve compensated.
0:09:47 And it looks different for every single person.
0:09:49 So that’s why the report is very valuable
0:09:51 ’cause for that person,
0:09:54 they know exactly which muscles are the ones
0:09:55 that they really need to target.
0:09:57 Both the ones that they already thought
0:10:00 maybe were an issue, but then all the other ones
0:10:03 that showed up and they didn’t really realize.
0:10:04 – And so in the case of this soccer player,
0:10:09 was it like one particular quadricep on one side
0:10:10 that was like the core thing
0:10:12 and you could figure out which one it was?
0:10:13 – There was a few muscles.
0:10:14 It wasn’t just that.
0:10:18 I think there were at least one calf muscle
0:10:21 and then some in the, especially in the deep hip,
0:10:23 those were impacted.
0:10:26 So yeah, it kind of shows up everywhere.
0:10:29 – And so you have this essentially diagnosis, right?
0:10:33 A very sort of fine-grained kind of diagnosis.
0:10:35 Do you also have a prescription?
0:10:39 Do you have sort of particular kinds of training
0:10:41 to address these very fine-grained things
0:10:44 or do you leave that to the trainers or whoever?
0:10:46 – We leave that to the trainers
0:10:48 because I think that it’s also important
0:10:51 to have all the other information about the athlete.
0:10:53 We’re not arguing that it replaces everything else.
0:10:56 And people pair it with lots of different
0:10:58 other types of measurements,
0:11:02 depending on the application or in the setting.
0:11:04 Like some people pair it with,
0:11:08 let’s say metrics of jump performance.
0:11:10 I’m shifting over to basketball here,
0:11:12 but that’s just one that came to mind
0:11:13 where you can look at the asymmetry
0:11:16 about how an athlete jumps,
0:11:17 but then you can also compare it
0:11:20 to the asymmetry of their muscles and get some insight.
0:11:24 So it definitely plugs in with a lot of other things.
0:11:29 – And to what extent can trainers or strength coaches
0:11:36 develop programs that are sufficiently kind of fine-grained
0:11:39 to match the kind of fine-grained findings you’re having, right?
0:11:43 Like for example, if you find, as I understand you did,
0:11:46 that a soccer player has one particular quadricep
0:11:49 that is weak, like are there workouts
0:11:52 that target a single quadricep and not the others?
0:11:53 – Yep, there are.
0:11:55 – That’s cool, for whichever quadricep you just,
0:11:58 like just for fun, give me an example.
0:12:01 – You know, one way that it’s very simple
0:12:03 is using something called biofeedback.
0:12:09 So you can measure whether you use something called EMG,
0:12:12 which is a way to measure how much
0:12:13 electrical activity is a muscle.
0:12:16 And then you can see which muscles you’re using
0:12:17 for a given task.
0:12:19 So if you give people the feedback
0:12:22 of which of those muscles they’re using
0:12:24 and say, “Oh, no, you’re not using this one,
0:12:27 “use this one more,” that actually works very effectively.
0:12:29 – Oh, really?
0:12:32 So you can basically use your brain
0:12:33 if you’re getting the feedback
0:12:36 to focus on which quadricep you’re training.
0:12:39 – Yeah, and there’s other ways you can give the feedback
0:12:41 in other different ways, but yeah,
0:12:43 our brains are very good at that
0:12:46 once they get feedback, they’re very good at learning.
0:12:47 – That’s cool, especially somehow
0:12:49 to think of with elite athletes, right?
0:12:52 ‘Cause they’re already presumably like super dialed in
0:12:54 in terms of like the relationship between their brain
0:12:56 and their body at this very elite level.
0:12:57 – Exactly.
0:12:59 Yeah, the other, I was gonna mention,
0:13:04 a lot of players and teams use this,
0:13:06 not just one time, but over time.
0:13:10 So they’ll get a scan, figure out a plan,
0:13:14 work on that for maybe three months or six months,
0:13:15 and then do another scan
0:13:18 and see how things are progressing and adjust accordingly.
0:13:21 So that’s definitely another way to,
0:13:23 in the long term, see if what they’re doing
0:13:27 is resulting in the change that they’re hoping to see.
0:13:29 – So what happened with that soccer player
0:13:33 who had the weak quadricep and other related troubles?
0:13:35 – Yeah, no, I think she’s doing great,
0:13:40 like staying healthy and getting ready.
0:13:42 – Yeah, so I know you can’t tell us her name,
0:13:44 but will we see her in the Olympics this summer?
0:13:45 – Yes.
0:13:46 – Great.
0:13:48 So as you were talking about that,
0:13:50 I mean, there was a moment where it was like,
0:13:52 okay, the athlete goes and gets the MRI,
0:13:56 and then you get the scan, you get the scan,
0:13:58 and then you said like, you crunch through the numbers,
0:13:59 and then you make the report.
0:14:02 Presumably, you crunching through the numbers
0:14:05 is like the result of many, many years of work
0:14:08 and kind of the core of what your company does.
0:14:10 So I wanna talk a little bit more about that
0:14:13 and kind of how you got here.
0:14:16 How did you come to start the company?
0:14:20 – How long do I have?
0:14:24 – A while, I mean, it’s not the radio, it’s a podcast.
0:14:26 – I’m warning you, I am a professor too,
0:14:27 so I can go on.
0:14:28 – Let me ask you this,
0:14:31 what was the moment when you decided to start the company?
0:14:36 – I can give you one moment, and then we can–
0:14:37 – Yeah, let’s try a couple of moments, sure.
0:14:39 – Yeah, let’s try a couple of moments in time.
0:14:44 – That’s what a good story is, like three moments in time.
0:14:46 – Right, right, three, not four.
0:14:49 – Four is a tricky number, we could do five or we could do one.
0:14:51 – Right, has to be odd, I think.
0:14:52 – Odd, I think, is better.
0:14:53 – Yeah, yeah.
0:14:55 So I’m a professor, I run a lab,
0:14:58 and for my entire career,
0:15:02 I’ve been fascinated with muscle and how it works.
0:15:06 And fascinated by something that we all,
0:15:10 in the muscle field, we call form function relationships.
0:15:13 So the idea that the way a muscle is shaped
0:15:16 and the way it’s structured and how big it is
0:15:20 influences how well it works, or how well it functions,
0:15:22 how strong it is, how well it behaves.
0:15:24 And there’s a lot to there, and there’s a lot of nuance,
0:15:27 and that’s like, I’ve spent a career studying that
0:15:29 in lots of different ways.
0:15:31 So I’ve always been interested in quantifying muscle
0:15:34 and figuring out how that influences how it works.
0:15:36 And both in healthy people or in athletes,
0:15:39 and also in different patient populations,
0:15:42 different, in particular, I have an interest
0:15:44 in movement disorders.
0:15:48 So neuromuscular diseases that lead to impairments
0:15:51 and mobility and movement ability.
0:15:55 So one of the light bulb moments for this
0:15:58 was the fact that I’d been using MRI
0:16:02 to study muscle in my research for a long time.
0:16:04 It’s kind of a ubiquitous tool,
0:16:07 or like often used tool in research.
0:16:11 But I was struck by the fact that
0:16:16 I was hearing from, in particular, a surgeon collaborator,
0:16:19 and the surgeon was telling me about his work
0:16:22 and helping kids with cerebral palsy
0:16:25 improve their movement, where they had hindered movement,
0:16:30 and largely because their muscles are impacted.
0:16:33 Not only do they have an impaired ability
0:16:36 to control their muscles, but their muscles end up
0:16:39 with impairments in their structure or their form,
0:16:41 which then influences how well they work.
0:16:43 So surgeons have to go in and do surgeries
0:16:45 to try to change that.
0:16:48 They do things like modify tendons
0:16:50 to try to make muscles less stiff,
0:16:55 or they transfer muscles to make them do a new thing.
0:16:58 But one of the big tricky parts is that oftentimes,
0:17:00 some of those muscles are very weak.
0:17:03 So if they choose the wrong muscle,
0:17:05 then they’ll make a weak muscle even weaker.
0:17:07 And that’s catastrophic.
0:17:10 So it’s a very fine line that a surgeon has to figure out,
0:17:11 and they have to go in.
0:17:14 We just talked about there’s 35 muscles in each leg.
0:17:16 So which of the muscles are the ones
0:17:20 that should be operated on, and which ones should be avoided?
0:17:25 And so what my collaborator, a guy named Dr. Mark Abel,
0:17:29 fantastic surgeon, he was telling me,
0:17:30 “Yeah, like it’s very hard.”
0:17:32 And I don’t, he didn’t have a way to see that.
0:17:35 All that he could do is look from the outside.
0:17:37 No technology could give him the information
0:17:39 he needed to figure out which muscles
0:17:41 he should focus on and which ones to avoid.
0:17:43 – Because it’s not obvious by looking,
0:17:46 what’s a strong muscle and what’s a weak one?
0:17:46 – Yeah.
0:17:49 – I guess that’s surprising to me on some level.
0:17:53 Like, I don’t, I’ve never thought about it,
0:17:56 but naively I would think you could look at a muscle
0:18:00 and say it looks strong or it looks weak, not so.
0:18:02 – Right, ’cause you just see it from the surface.
0:18:04 You don’t see it on the inside.
0:18:08 – And the other challenge is for every joint,
0:18:09 there’s many muscles.
0:18:12 So like we just said, the quadriceps has four muscles.
0:18:14 On the back of the leg, hamstrings,
0:18:17 there’s three hamstrings muscles.
0:18:19 There’s other muscles that are in the thigh.
0:18:22 So you’re just seeing what’s an impairment
0:18:24 in the overall movement, let’s say, of a joint,
0:18:27 but then there could be many muscles or combinations
0:18:28 of muscles that are leading to that.
0:18:31 And you don’t know when you look from the outside.
0:18:34 Our body is designed that way actually to be somewhat,
0:18:35 we call it redundant.
0:18:37 We have more muscles than we need probably,
0:18:40 but if you think about imbalances,
0:18:44 then any one of those muscles could create some trouble.
0:18:47 – So okay, so the surgeon describes this problem
0:18:50 he’s having, and then what do you do?
0:18:53 – So then I was thinking, well, you know,
0:18:56 that’s the information that we generate all the time
0:18:58 when we’re doing our research.
0:19:03 We take MRIs, we quantify, we identify muscles,
0:19:05 we create three-dimensional models of the muscles,
0:19:08 we figure out how they’re working from that.
0:19:11 But I was struck by the fact that none of that
0:19:13 was something that a clinician can use.
0:19:17 Despite the fact that MRI is obviously ubiquitous
0:19:19 in healthcare, right?
0:19:22 You can’t go to a hospital without finding multiple MRIs,
0:19:24 but there’s no way to use those MRIs
0:19:27 in the way that I was using them for my research.
0:19:30 And I thought, well, that’s too bad
0:19:33 because this would be very useful to the surgeon
0:19:37 in figuring out how to treat these patients.
0:19:40 So that was one light bulb at the beginning.
0:19:42 So a lot of it was figuring out
0:19:44 how to take something that we use in research
0:19:48 for these very specific targeted basic science questions
0:19:52 and turn it into something that is useful clinically.
0:19:54 – And again, this is like my ignorance.
0:19:56 Like I might have thought, well,
0:20:00 you could just do an MRI and see how big
0:20:03 or not big the muscles are and infer from how big
0:20:07 or not big the muscles are, how strong or not strong they are.
0:20:10 And that sounds straightforward,
0:20:13 but clearly it’s not.
0:20:15 Like why is it harder than that?
0:20:17 – So a couple of reasons.
0:20:20 One is going, taking the MRI pictures
0:20:22 and figuring out how big the muscles
0:20:25 is a very challenging problem.
0:20:28 So in order to accurately get how big the muscles are,
0:20:33 you have to essentially generate its shape in three dimensions.
0:20:35 So you have to get the whole length of the muscle.
0:20:39 And so you do that off of multiple MRI pictures.
0:20:43 So the MRI essentially kind of takes pictures
0:20:47 through the body at multiple different slices, we call them,
0:20:51 going from the abdomen all the way down to the feet,
0:20:54 sort of going cross-sectionally, we call it.
0:20:58 And so we usually have over 200 images like that.
0:21:01 So in each image, you have to find each muscle.
0:21:03 And so for any given image,
0:21:08 there’s probably at least 15 muscles or more.
0:21:11 – So it wasn’t like you could just push the like,
0:21:14 show me the muscles button on the MRI
0:21:15 and it would show you the muscles.
0:21:16 Like nobody had done that
0:21:18 and there was no obvious way to do it.
0:21:22 Certainly not for a surgeon ordering a standard MRI,
0:21:24 just didn’t exist.
0:21:25 – It did not exist.
0:21:30 – So you realize this, what happens?
0:21:32 How do you make it happen?
0:21:35 – So one of our first tasks was to figure out
0:21:37 how to get many muscles.
0:21:41 So one of the things that we had done on the research side
0:21:43 is really focus on a couple of muscles.
0:21:46 But I knew for this application, that wasn’t gonna work,
0:21:49 we have to be able to identify any muscle.
0:21:51 That was really the problem,
0:21:53 is that like you don’t know which one’s the problem.
0:21:54 So you don’t know which one to look at.
0:21:56 So you gotta look at all of them.
0:21:59 And then the next task was to figure out all those muscles
0:22:03 and figure out a process to go from the,
0:22:06 identifying each and every muscle and each and every image.
0:22:08 So it is called developing an atlas.
0:22:11 – And is that an AI problem?
0:22:15 – So now we have an AI and it’s the type of AI,
0:22:16 a supervised learning,
0:22:19 where you essentially train the computer
0:22:22 to do what the person would do.
0:22:23 But in order to do that,
0:22:25 you need to do what the person would do first.
0:22:31 And so we did that all manually at first.
0:22:34 In order to generate one of these reports,
0:22:39 at first it took us about 50 hours per person.
0:22:42 – Just going through image after image after image
0:22:45 and saying this is this muscle, that is that muscle?
0:22:46 – Exactly.
0:22:47 So we needed to develop that.
0:22:50 But the other piece we needed is this,
0:22:53 back to this normative database I talked about.
0:22:56 Because if I just told you how big your muscle
0:22:58 is in milliliters in volume,
0:23:00 what are you gonna do with that information?
0:23:01 Like, oh, great.
0:23:03 – And nobody knew, and it’s interesting,
0:23:05 it’s one of those things you always think,
0:23:06 oh, surely there’s some data in the world
0:23:07 that everybody knows X.
0:23:10 But so you’re saying nobody knew what was the kind of
0:23:13 median size of a particular quadricep
0:23:16 for whatever, a healthy 12 year old boy or whatever.
0:23:18 Nobody knew that at that time?
0:23:21 – Well, all of the information up until then,
0:23:25 for the most part, was based on dissecting cadavers.
0:23:30 Was based on taking cadavers and dissecting muscles,
0:23:33 weighing the muscles.
0:23:36 And one of the big challenges with that is usually
0:23:38 cadavers are older adults.
0:23:41 And so they’re not really representative
0:23:45 of a younger, healthy population.
0:23:47 And I will tell you at that time,
0:23:49 that that was a lot of work.
0:23:52 And I had people saying, like, why are you doing that?
0:23:55 Like, that seems like a waste of time.
0:23:56 That’s crazy.
0:24:01 I had this vision and I trusted that it was gonna
0:24:02 turn into something at least useful
0:24:04 to the research community.
0:24:06 And I’m thankful that we stuck with it.
0:24:10 – There’s lots more to come on the show,
0:24:12 including, but not limited to,
0:24:14 the work Sylvia and her colleagues
0:24:16 are doing with major league pitchers,
0:24:17 college football players,
0:24:20 and patients with degenerative muscle disease.
0:24:34 Sylvia and her colleagues trained an AI model
0:24:38 to do what had previously taken a human 50 hours
0:24:40 for every person who got scanned.
0:24:43 And they expanded from working with patients
0:24:47 with cerebral palsy to working with elite athletes.
0:24:50 Today, their clients include not just Olympic athletes,
0:24:54 but teams in the NBA and the Premier League.
0:24:56 Also, she told me they’re working on a project
0:24:58 with major league baseball.
0:25:02 – Yeah, so we’re working with the MLB studying pitchers
0:25:04 and we’re getting essentially a normative database
0:25:05 whole body scan of pitchers.
0:25:08 – And is that partly because like,
0:25:10 pitchers mess up their arms so badly?
0:25:12 Is that kind of the motivation there?
0:25:16 – Yes, there’s definitely a lot of issue
0:25:17 with injury and surgery.
0:25:21 And so the idea here is that by taking these scans,
0:25:25 we can really figure out where there might be weaknesses
0:25:30 and sort of potential areas for mitigating the injuries.
0:25:33 – So when you do work for a whole team,
0:25:37 like say the bowls, a basketball team, an NBA team,
0:25:40 like what’s the nature of that work?
0:25:42 What do you do for a team like that?
0:25:46 – Yeah, they will do a baseline of the whole team.
0:25:51 – And they basically tailor the athlete’s training,
0:25:54 presumably strength training in particular,
0:25:56 like on a muscle by muscle basis
0:25:59 based on the reports that you’re sending them.
0:26:00 – Correct, yeah, yeah.
0:26:04 – And I mean, you can imagine like better performance
0:26:06 being one outcome.
0:26:10 Reduced risk of injury seems plausible, right?
0:26:13 Like it seems obvious that like a big asymmetry
0:26:15 could make you more likely to be injured.
0:26:17 I mean, are you at a point now
0:26:20 where you can predict the risk of injury?
0:26:24 – That’s like a whole can of worms.
0:26:26 I won’t say that– – Let’s open that can of worms
0:26:26 for a sec. – Yes, let’s open it.
0:26:28 – I mean, is that interesting to you
0:26:30 or is that like too much or yeah?
0:26:33 – No, no, no, this is something we think about a lot.
0:26:38 And let me, I wanna, so first I’ll tell you
0:26:39 why it’s a can of worms.
0:26:42 But then I’ll tell you what project that we’re doing.
0:26:43 – Then tell me about the can.
0:26:44 We’ll look at it from the outside.
0:26:46 – So from the can of, like there’s a lot of technologies
0:26:50 out there that will say that they’re predicting injury risk.
0:26:51 They’ll give you numbers
0:26:54 and they’re just not based on anything.
0:26:59 And so I don’t know, there’s a lot of–
0:27:02 – So it’s the land of specious claims, yeah.
0:27:03 – Yeah, yeah, yeah.
0:27:05 So that’s not what we’re about.
0:27:09 We’re about like providing actual things that matter.
0:27:14 And so the question is like, can you do these muscle scans?
0:27:20 Do they correlate with injury likelihood in some way?
0:27:23 And so we actually have a project
0:27:24 to address that very question.
0:27:26 It’s actually funded by the NFL.
0:27:29 We’re actually in that project.
0:27:32 We’re working with college teams.
0:27:33 – College football, college football.
0:27:34 – College football teams, yeah.
0:27:39 Baselining entire rosters at the beginning of the season.
0:27:42 And then tracking hamstring injuries.
0:27:45 And then if an athlete gets injured,
0:27:47 they come back for a scan at the time of injury
0:27:50 and then at return to sport.
0:27:54 And so one of our questions is based on the baseline scan,
0:27:58 can we predict who’s more likely to get an injury,
0:28:00 an initial injury, index injury?
0:28:02 And then the secondary question is,
0:28:05 can we predict who will be re-injured?
0:28:08 Saying that we often people pair it with other things.
0:28:11 In this project, we’re always also doing that.
0:28:13 Each athlete is getting an assessment
0:28:15 of their sprint mechanics.
0:28:18 So kind of the biomechanics of how they run.
0:28:22 And then also assessment of their strength,
0:28:23 of their hamstring muscle.
0:28:25 So kind of measured strength.
0:28:27 Obviously you can’t do that when they have an injury,
0:28:30 of course, but you can when they’re healthy.
0:28:33 – That biomechanics piece seems like something
0:28:36 that has been developing in parallel with your work.
0:28:39 Also driven by computer vision, right?
0:28:44 The like markerless motion capture seems like a big world
0:28:49 that overlaps with your world some.
0:28:54 So tell me about your work with female athletes
0:28:59 versus male athletes and how that plays a role.
0:29:01 – Yeah, I will say probably the biggest thing
0:29:05 that we’ve been focused on is making sure
0:29:07 that our data addresses that.
0:29:12 So our normative database is separated by sex.
0:29:18 And it is different because women aren’t small men, right?
0:29:22 So it’s important that we have that basis to compare
0:29:27 that’s like for women and not comparing to some average
0:29:29 or primarily male data set.
0:29:31 So that’s one huge important thing
0:29:35 is that it’s compared to the normative values
0:29:37 for the female population.
0:29:42 And then in terms of like working with the female athletes,
0:29:46 I think one of the big ones is really just ability
0:29:50 to personalize and provide this
0:29:55 like really accurate detailed assessment of their bodies.
0:29:58 And a lot of the knowledge
0:30:02 about like appropriate body composition
0:30:04 historically has been based on studies in men.
0:30:08 And but then we’re applying them to women
0:30:11 and making us feel really bad about ourselves.
0:30:15 So really motivated to move away from that
0:30:18 and sort of acknowledge the muscular physiology
0:30:20 and anatomy of the female
0:30:25 and also the female athlete to really understand that.
0:30:28 I think one thing obviously that we’ve seen
0:30:32 is ACL injuries are more common in women than men
0:30:37 and examining how these like recovery profiles look
0:30:40 and how they differ between men and women.
0:30:41 That’s something that we’re observing
0:30:43 and seeing how those things shake out.
0:30:47 But we’re motivated by really providing that information
0:30:48 that’s specific to women.
0:30:53 – So what are some of the non sports things
0:30:55 you’re working on, things you’re trying to figure out?
0:31:00 – Yeah, I mean, one that I’m really interested in
0:31:03 is this area that we’re applying to
0:31:06 in clinical trials for muscle disease.
0:31:10 So we’ve been working in a specific muscle disease
0:31:15 called fascioscabular humeral muscular dystrophy, FSHD,
0:31:19 which is a slowly progressing muscle disease,
0:31:20 genetic in basis.
0:31:25 And so eventually people with FSHD need a wheelchair
0:31:28 just life is very difficult.
0:31:30 And so it’s pretty devastating.
0:31:31 But the other exciting thing
0:31:33 is there are some new treatments out there,
0:31:36 some in particular gene therapies coming online.
0:31:39 And now the challenge is, do they work?
0:31:42 ‘Cause the problem is in these diseases
0:31:44 ’cause they’re pretty slowly progressing.
0:31:49 If you wanna see if a drug is helping somebody,
0:31:52 it’s very hard to see that in a slowly progressing disease.
0:31:55 – The clinical manifestations are hard to pick up.
0:31:57 – They are. – If it makes your muscles
0:32:01 shrink more slowly, it’s gonna be hard to see.
0:32:02 – It’s very hard to see,
0:32:04 especially from like rudimentary measures.
0:32:07 But with the MRIs, we’ve been able to provide
0:32:09 this really detailed insight
0:32:11 about the disease state of each muscle
0:32:14 and how it’s progressing over time.
0:32:17 And so one of our goals is to really lean in on this
0:32:21 and help figure out exactly how people should look
0:32:23 at all this data and figure out
0:32:25 if a drug is working or not.
0:32:28 It’s really profoundly important because without that,
0:32:30 these clinical trials just won’t move forward.
0:32:37 – What else are you sort of still trying to figure out?
0:32:40 – So we talked about predicting injury,
0:32:44 but having all the data needed to show,
0:32:47 like if your scan looks like this and if you do this,
0:32:50 you will be able to improve your jump height by that.
0:32:53 – Yeah, you’ll be able to throw a fastball
0:32:54 two miles an hour faster.
0:32:57 Like that would be wildly valuable.
0:33:00 – That would be very, and we do have data in our research.
0:33:05 We were able to show that these muscle scores
0:33:07 correlate with performance metrics,
0:33:09 such as jump height and speed.
0:33:11 So we for sure see that.
0:33:13 The question is then the spin on like observing
0:33:15 how that plays out.
0:33:19 Like if you then strengthen the appropriate muscles,
0:33:21 how much faster do you get?
0:33:23 And you just need more and more data
0:33:26 to really like to go after that.
0:33:28 But that’s one thing that I’m fascinated by.
0:33:30 One of the other interesting ones,
0:33:32 I’ll go, can I go off on a tangent?
0:33:33 – Any, anything you want.
0:33:36 – One of our research partners
0:33:39 that’s interested in how muscles adapt to strength training
0:33:42 and different interventions and what influences that
0:33:45 had a really interesting finding
0:33:48 that I think is quite profound, but also obvious.
0:33:51 So everybody, if they’re targeted training,
0:33:53 their quadriceps and hamstrings,
0:33:56 those muscles got bigger.
0:33:57 That makes sense.
0:33:59 But in a fair number of the people,
0:34:01 some other muscles got smaller.
0:34:06 And then he had done some controlling
0:34:10 and documentation of nutrition intake.
0:34:14 And he found that people that had higher caloric
0:34:19 and protein intake had less of that effect.
0:34:21 – So all the gym bros telling you
0:34:25 to eat a lot of protein are validated by this guy’s study.
0:34:28 – Yeah, but it’s not necessarily to make that muscle
0:34:29 that you’re working bigger.
0:34:30 So you don’t lose the other muscles.
0:34:34 – That’s a good one.
0:34:37 And was he using your scans to figure that out?
0:34:38 – Yeah, he was using our scans.
0:34:41 And the thing that was cool is that normally
0:34:43 in research, you wouldn’t bother looking
0:34:44 at those other muscles.
0:34:46 You would just look at the ones that were targeted.
0:34:48 ‘Cause those are the ones you just think about.
0:34:52 But by getting the entire extent of all the muscles,
0:34:55 you see these impacts that you wouldn’t necessarily
0:34:56 have noticed.
0:34:58 – Like he wasn’t even looking for it.
0:34:59 – Yeah, it’s quite profound
0:35:01 because somebody’s strength training,
0:35:03 recovering from an injury,
0:35:08 that really means the nutritional element’s important
0:35:10 because you could be strengthening some muscles,
0:35:14 but weakening others if you’re not playing your cards
0:35:15 right there.
0:35:20 – We’ll be back in a minute with the lightning round.
0:35:33 I wanna finish with the lightning round.
0:35:34 Won’t take too long.
0:35:35 It’ll be fun.
0:35:36 – Okay.
0:35:37 – What?
0:35:38 – I don’t know what that is.
0:35:40 – Well, you’ll find out right now.
0:35:43 – Does this have to be fast?
0:35:43 – Nope.
0:35:44 (laughing)
0:35:46 – I could call it the random round.
0:35:48 – Okay, I like that, random.
0:35:52 – Have you scanned yourself?
0:35:53 – Yes.
0:35:54 Multiple times.
0:35:55 – Of course.
0:35:56 – Of course.
0:35:58 – I guess what kind of thing are you talking about?
0:35:59 I’ve probably been in an MRI machine,
0:36:01 I don’t know, maybe a hundred times,
0:36:02 like a lot of times.
0:36:03 – It’s not radiation, right?
0:36:04 It’s not like an x-ray.
0:36:05 You could do it every day if you want to.
0:36:07 – Yeah, as much as you want, yeah.
0:36:08 – What’d you learn?
0:36:11 – So I actually used it,
0:36:13 I’ve learned lots over the years,
0:36:15 but I will tell you one anecdote,
0:36:17 I have a hip replacement.
0:36:23 I have a genetic condition that leads to early arthritis.
0:36:27 And so I was, before I got my hip replacement,
0:36:28 I got a scan.
0:36:30 I knew I was getting weak,
0:36:33 but holy cow, was I really weak on that side?
0:36:38 What was profound was how weak my hip flexors were,
0:36:40 very weak.
0:36:42 And I think a lot of times people talk
0:36:44 about hip flexors being tight.
0:36:47 And that’s kind of what I thought was happening.
0:36:51 I felt pain and I felt like I was having a lot of tightness,
0:36:53 but it was actually weakness.
0:36:57 And they were like super small on both sides,
0:37:00 but really especially on the side that was affected.
0:37:04 So that was one thing that I worked on a lot.
0:37:06 – Did that genetic condition you have,
0:37:08 did that influence your work at all,
0:37:10 your decision to go into the field?
0:37:15 – I mean, like loosely maybe,
0:37:18 because my dad had the same thing,
0:37:20 which actually caused him to go blind.
0:37:21 – Oh goodness.
0:37:23 – It has like a multiple different issues.
0:37:26 And so I think that at a early age got me interested
0:37:31 in medicine and disabilities and helping people.
0:37:37 So that might be broadly speaking.
0:37:38 And I knew I had some eye problems.
0:37:40 I didn’t know the genetic thing.
0:37:42 We didn’t discover that till later, but.
0:37:43 – Huh, interesting.
0:37:48 What’s the most underrated muscle in the human body?
0:37:52 – That’s a hard one.
0:37:57 So I have a few favorite muscles.
0:37:58 – Okay, what’s your favorite muscle?
0:38:03 – Yeah, so the psoas muscle, psoas major, it’s a hip flexor.
0:38:05 But it also, it’s really cool.
0:38:09 It actually, it’s also a back muscle, lower back muscle.
0:38:11 So it attaches to the lumbar vertebrae,
0:38:15 but then it also crosses the front of your hip.
0:38:16 It’s really hard to find
0:38:19 because it’s like really back deep in your hip.
0:38:21 It goes right over your, your femur.
0:38:24 – So it goes like in the middle of your body kind of.
0:38:25 – Yeah, right in the middle.
0:38:26 It kind of connects everything.
0:38:29 Sort of connects your lower extremity
0:38:31 to the rest of your body in some ways.
0:38:34 – Okay, last one.
0:38:38 – Why do you hate astrophysicist Barbie?
0:38:41 – I don’t hate anything.
0:38:42 – I mean.
0:38:45 – Well, it’s too perfect.
0:38:50 It’s kind of like this, you know, idea that like,
0:38:55 oh, you know, you can inspire girls to go into science
0:38:56 by showing them that Barbie does too.
0:38:59 But Barbie is like fictitious.
0:39:00 So it kind of tells you that
0:39:06 like promotes the idea of perfectionism in society,
0:39:09 but definitely in girls.
0:39:11 And, you know, what we really want to promote
0:39:14 is almost the opposite of that, is taking risks
0:39:18 and not worrying about being perfect
0:39:21 and just doing something that matters to you.
0:39:23 So yeah, I don’t know.
0:39:24 I mean, I don’t hate it.
0:39:26 I had Barbies when I was a kid, but I just,
0:39:28 it kind of like, so it’s something that-
0:39:29 – You wrote a whole column.
0:39:32 You wrote a whole column in the newspaper about-
0:39:33 – I did.
0:39:34 – About why it all worked out.
0:39:35 – Yeah, wasn’t that cool?
0:39:36 I was very proud of that.
0:39:39 Yeah, no, and really that, you know,
0:39:43 the astrophysicist part honestly was more of a hook.
0:39:45 The article I had written already
0:39:48 before that astrophysicist Barbie came to be,
0:39:51 it was about the issue of perfectionism
0:39:56 and how that dissuades girls to go into STEM and research.
0:40:02 – Well, a good hook is important, and you found one.
0:40:03 – Yeah, it was perfect, yeah.
0:40:10 – Sylvia Blemker is a professor at the University of Virginia
0:40:14 and the co-founder of Springbok Analytics.
0:40:16 Next week on What’s Your Problem,
0:40:17 I’ll be talking to Jimmy Buffy.
0:40:21 He is using AI to bring the insights of biomechanics
0:40:24 to professional athletes.
0:40:26 Jimmy told me that before the advent of AI,
0:40:29 when biomechanics experts tried to work with athletes,
0:40:32 it could be somewhat awkward.
0:40:34 – So you’ve got like a picture in his underwear
0:40:36 with a bunch of little metal balls.
0:40:38 They’re like, just pitch like you always pitch.
0:40:40 – Right, and the state of the art for tracking it
0:40:42 was an awful experience
0:40:45 for the people you were trying to track.
0:40:46 – So what changes?
0:40:49 – The big inflection point was computer vision,
0:40:51 basically using artificial intelligence
0:40:55 to identify where those joints are in a camera image
0:41:00 rather than needing to paste those reflective markers.
0:41:03 – Today’s show was produced by Gabriel Hunter-Chang,
0:41:05 edited by Lydia Jean-Cott
0:41:08 and engineered by Sara Bouguere.
0:41:12 You can email us at problem@pushkin.fm.
0:41:14 I’m Jacob Goldstein, we’ll be back next week
0:41:16 with another episode of What’s Your Problem.
0:41:19 (upbeat music)
0:41:21 (upbeat music)
0:41:25 (upbeat music)
0:41:27 (upbeat music)
0:41:37 [BLANK_AUDIO]
0:00:05 (crowd cheering)
0:00:09 – Pushkin.
0:00:19 What’s one surprising thing your work has taught you
0:00:22 about elite athletes?
0:00:24 – I never thought I would see muscles
0:00:26 that were so developed.
0:00:28 (laughing)
0:00:30 They broke our scale.
0:00:30 – Wow.
0:00:32 Yeah.
0:00:33 – Like it was just too big.
0:00:35 The machine, the AI couldn’t figure out what it is.
0:00:37 – Well, no, the AI found it,
0:00:40 but we’re like our kind of rating system.
0:00:40 – Wow.
0:00:43 Was there a particular athlete or a particular sport
0:00:44 or a particular muscle?
0:00:46 What muscle broke the scale?
0:00:50 – The gluteus maximus breaks it at fair amount.
0:00:51 – No kidding.
0:00:52 – Yeah, yeah.
0:00:53 – Fantastic.
0:00:53 – Yes.
0:00:54 It’s a pain in my butt.
0:00:56 (laughing)
0:00:58 Like ’cause it’s too big?
0:01:01 – Yeah, yeah, it’s just so big.
0:01:03 (upbeat music)
0:01:09 – I’m Jacob Goldstein and this is What’s Your Problem.
0:01:11 This month, a bunch of Pushkin podcasts
0:01:15 are coming out with Olympics-inspired shows.
0:01:18 Revisionist History has a series about America’s decision
0:01:23 to participate in Hitler’s Berlin Olympics in 1936.
0:01:25 The Happiness Lab has an interview with a coach
0:01:27 who coaches coaches.
0:01:29 And here on What’s Your Problem,
0:01:31 we’re gonna be talking with people
0:01:34 who are working at the frontiers of technology
0:01:37 to help elite athletes perform better.
0:01:42 For example, today my guest is Sylvia Blemkert.
0:01:45 She’s a professor of biomechanical engineering
0:01:47 at the University of Virginia.
0:01:48 And she’s the co-founder of a company
0:01:51 called Springbok Analytics.
0:01:53 Sylvia’s problem is this.
0:01:56 How do you combine MRI scans and artificial intelligence
0:02:00 to generate new insights that can help both elite athletes
0:02:04 and people suffering from diseases that affect the muscles?
0:02:07 Springbok’s clients include medical researchers,
0:02:09 Olympic athletes, major league baseball,
0:02:13 and several professional basketball and soccer teams.
0:02:15 You’ll hear about all that on the show.
0:02:17 But first, we’re gonna pick up
0:02:19 where we left off in the conversation.
0:02:23 We were discussing the extraordinarily large muscles
0:02:25 of elite athletes.
0:02:29 And then Sylvia told me something even more surprising.
0:02:32 – The other thing is that they have some tiny muscles too.
0:02:36 – Huh, like smaller than a normal person’s muscle.
0:02:38 – Yes, much smaller.
0:02:40 They put their muscle where they need it.
0:02:41 – What’s an example?
0:02:44 Like what muscle is tiny and what kind of athlete?
0:02:48 – Calf muscles are small in most fast athletes.
0:02:54 Like you look at a sprinter or like a running back.
0:02:56 – It’s just all quad, no calf?
0:02:59 – All like thigh, no calf.
0:03:00 Yeah, thigh and hip.
0:03:03 It kind of makes sense because if you’re trying to run fast,
0:03:04 you wouldn’t want to put a lot of mass
0:03:06 like at the end of your leg.
0:03:09 It’s like adds a lot of inertia to like move your leg.
0:03:12 Because the muscles are important for sprinting.
0:03:15 That’s the interesting thing, but they just don’t.
0:03:16 They’re small, they’re very small.
0:03:18 – Uh-huh, uh-huh.
0:03:20 So I’m particularly interested at this moment
0:03:24 in the sports piece of what you do.
0:03:25 – Uh-huh, good.
0:03:28 – I’m curious, by the way,
0:03:33 do you work with any Olympic teams or Olympic athletes?
0:03:35 – Yeah, yeah, we’ve actually been working
0:03:38 with several different Olympic athletes.
0:03:41 The ones that probably that come to mind most
0:03:46 are multiple players on the U.S. women’s national soccer team.
0:03:46 – Oh, cool.
0:03:50 Tell me like, tell me the story of that work.
0:03:53 So they came to you, what did they,
0:03:54 what did they want when they came to you?
0:03:55 Like how did that begin?
0:04:00 – They came to us along with their team.
0:04:02 So the technology we provide, you know,
0:04:04 an athlete could understand it,
0:04:07 but really with their team to help them figure out
0:04:10 how to keep athletes healthy.
0:04:12 – So what did they say?
0:04:14 What did they say when they came to you?
0:04:19 – So, for example, one athlete that’s coming to mind
0:04:23 had a known imbalance, side to side,
0:04:26 that based on a history of injury.
0:04:28 And they really wanted to know
0:04:29 where that imbalance was coming from.
0:04:32 – So the woman had hurt one of her legs
0:04:35 and that leg was, even after she came back,
0:04:38 that leg was weaker essentially than the other.
0:04:40 I mean, is that the sort of gross, you know, macro view?
0:04:42 – Yeah, that’s a fair way to say, yeah, exactly.
0:04:44 That’s a nice way to put it, yeah.
0:04:46 And they wanted to sort of find her like,
0:04:48 okay, but we can see that,
0:04:50 but what’s going on on the inside?
0:04:52 Like muscle by muscle, tell us that.
0:04:53 – Yes, exactly.
0:04:56 That’s precisely what we do, we go on the inside.
0:04:58 ‘Cause on the outside,
0:05:03 you see perhaps that her knee extensor or quads
0:05:04 seem weaker on one side than the other,
0:05:09 but there’s four quads, quadriceps, four muscles.
0:05:11 And so it’s not clear which of those muscles
0:05:14 are actually the culprit for that imbalance
0:05:15 and in what way.
0:05:17 – Good, so this is their question
0:05:20 and then what happens next?
0:05:23 – So the first step is an MRI scan.
0:05:28 And so with these athletes or teams,
0:05:33 we have ways to connect them with an MRI machine,
0:05:36 whether it be through an imaging center
0:05:37 that they partner with
0:05:42 or we’ve even actually brought MRI mobile trucks to sites
0:05:44 to make it easier for them.
0:05:46 – It’s like the players run off the field
0:05:49 and get an MRI and go back and keep playing, yeah.
0:05:51 – Yeah, kind of, yeah.
0:05:53 It helps just with the timing of things.
0:05:56 But so first we connect them there.
0:05:58 So it takes about 10 minutes.
0:06:01 Then they send those pictures up into the cloud
0:06:04 into our server.
0:06:06 And then we crunch through it
0:06:10 and then we send back a report on their muscles.
0:06:13 We also have what we call interactive viewer
0:06:17 and it’s presented in the form of a 3D model,
0:06:18 three-dimensional model.
0:06:21 So you actually see your own legs,
0:06:23 the muscles and bones, your own muscles and bones
0:06:27 that we’ve identified from the images,
0:06:29 going through a process called segmentation
0:06:32 where we find all the muscles and bones
0:06:33 and then we reconstruct them.
0:06:36 So it’s kind of like a digital twin of that person
0:06:38 that they can see on their computer.
0:06:41 And so along with it are all these metrics
0:06:44 that helps them understand their balance,
0:06:48 the development or strength of the muscles
0:06:50 and the health of the muscles.
0:06:52 – So tell me about this report they get.
0:06:54 Like what does it say?
0:06:58 – So the basis of that is actually a lot of research
0:06:59 that we did over many years
0:07:03 because you need to understand
0:07:05 where somebody falls relative to a normal,
0:07:07 essentially to give them,
0:07:10 essentially we have a scoring system for the muscles.
0:07:12 And that’s based on comparing
0:07:16 with a large data set of healthy individuals.
0:07:19 And so we know for a given person
0:07:24 based on their sex, age, height and weight,
0:07:27 how big we expect all the muscles to be.
0:07:29 And that’s through a lot of previous research.
0:07:32 So then we can say, okay, here’s where you land
0:07:36 each particular muscle compared to this,
0:07:38 what we call a normative database.
0:07:40 So we call it a spring box score.
0:07:43 – Do you do it for every muscle in the leg or?
0:07:44 – We do it.
0:07:46 So our primary product that we started with
0:07:51 was every muscle in the legs, essentially from belly to feet.
0:07:53 No muscle left behind.
0:07:54 They’re all important.
0:07:56 – How many muscles are there?
0:07:58 – 35 per leg.
0:08:00 So 70. – Okay, okay.
0:08:01 More than I would have guessed.
0:08:02 But what do I know?
0:08:03 – There’s a lot in there.
0:08:04 – Yeah, there’s a lot in there.
0:08:06 – They’re all important.
0:08:11 – Okay, so it’s basically how strong and healthy
0:08:16 is every one of those 70 muscles relative to baseline?
0:08:19 – And then the asymmetry comes
0:08:20 where you can compare side to side.
0:08:22 So for each of the 35 muscles
0:08:24 that they exist on each leg,
0:08:26 we can say which side is bigger,
0:08:28 which side is smaller, and by how much.
0:08:32 And then we also have normative values for that
0:08:34 because we’re all just slightly asymmetric, right?
0:08:37 – And presumably some muscles are more asymmetric
0:08:38 than others.
0:08:42 And so you wanna know kind of how asymmetric relative
0:08:45 to baseline is this particular pair of muscles.
0:08:46 – Exactly, yeah.
0:08:52 – And so in the case of this soccer player
0:08:56 who came to you who knew she had some kind of problem
0:08:57 with her quadriceps on one side
0:09:00 but didn’t know what was going on, what did you find?
0:09:03 – We found some imbalances,
0:09:06 actually not just in those muscles.
0:09:09 It turns out that it’s all connected.
0:09:14 So if you have a weakness or an imbalance
0:09:16 in one set of muscles, usually some other set of muscles
0:09:18 are compensating in some way.
0:09:21 – Well, it’s like when you mess up,
0:09:24 even if you’re just a recreational athlete, right?
0:09:25 Like if you mess up something,
0:09:27 you mess up your ankle, then you start walking funny
0:09:29 and then like your back hurts
0:09:30 ’cause you’re walking funny, right?
0:09:33 Like that is a very anecdotally apparent thing.
0:09:34 – Yeah, yeah, we all know that,
0:09:36 but it shows through in the scan.
0:09:39 But the thing is that it’s not very intuitive
0:09:43 from the outside which muscles have been affected
0:09:44 and how they’ve compensated.
0:09:47 And it looks different for every single person.
0:09:49 So that’s why the report is very valuable
0:09:51 ’cause for that person,
0:09:54 they know exactly which muscles are the ones
0:09:55 that they really need to target.
0:09:57 Both the ones that they already thought
0:10:00 maybe were an issue, but then all the other ones
0:10:03 that showed up and they didn’t really realize.
0:10:04 – And so in the case of this soccer player,
0:10:09 was it like one particular quadricep on one side
0:10:10 that was like the core thing
0:10:12 and you could figure out which one it was?
0:10:13 – There was a few muscles.
0:10:14 It wasn’t just that.
0:10:18 I think there were at least one calf muscle
0:10:21 and then some in the, especially in the deep hip,
0:10:23 those were impacted.
0:10:26 So yeah, it kind of shows up everywhere.
0:10:29 – And so you have this essentially diagnosis, right?
0:10:33 A very sort of fine-grained kind of diagnosis.
0:10:35 Do you also have a prescription?
0:10:39 Do you have sort of particular kinds of training
0:10:41 to address these very fine-grained things
0:10:44 or do you leave that to the trainers or whoever?
0:10:46 – We leave that to the trainers
0:10:48 because I think that it’s also important
0:10:51 to have all the other information about the athlete.
0:10:53 We’re not arguing that it replaces everything else.
0:10:56 And people pair it with lots of different
0:10:58 other types of measurements,
0:11:02 depending on the application or in the setting.
0:11:04 Like some people pair it with,
0:11:08 let’s say metrics of jump performance.
0:11:10 I’m shifting over to basketball here,
0:11:12 but that’s just one that came to mind
0:11:13 where you can look at the asymmetry
0:11:16 about how an athlete jumps,
0:11:17 but then you can also compare it
0:11:20 to the asymmetry of their muscles and get some insight.
0:11:24 So it definitely plugs in with a lot of other things.
0:11:29 – And to what extent can trainers or strength coaches
0:11:36 develop programs that are sufficiently kind of fine-grained
0:11:39 to match the kind of fine-grained findings you’re having, right?
0:11:43 Like for example, if you find, as I understand you did,
0:11:46 that a soccer player has one particular quadricep
0:11:49 that is weak, like are there workouts
0:11:52 that target a single quadricep and not the others?
0:11:53 – Yep, there are.
0:11:55 – That’s cool, for whichever quadricep you just,
0:11:58 like just for fun, give me an example.
0:12:01 – You know, one way that it’s very simple
0:12:03 is using something called biofeedback.
0:12:09 So you can measure whether you use something called EMG,
0:12:12 which is a way to measure how much
0:12:13 electrical activity is a muscle.
0:12:16 And then you can see which muscles you’re using
0:12:17 for a given task.
0:12:19 So if you give people the feedback
0:12:22 of which of those muscles they’re using
0:12:24 and say, “Oh, no, you’re not using this one,
0:12:27 “use this one more,” that actually works very effectively.
0:12:29 – Oh, really?
0:12:32 So you can basically use your brain
0:12:33 if you’re getting the feedback
0:12:36 to focus on which quadricep you’re training.
0:12:39 – Yeah, and there’s other ways you can give the feedback
0:12:41 in other different ways, but yeah,
0:12:43 our brains are very good at that
0:12:46 once they get feedback, they’re very good at learning.
0:12:47 – That’s cool, especially somehow
0:12:49 to think of with elite athletes, right?
0:12:52 ‘Cause they’re already presumably like super dialed in
0:12:54 in terms of like the relationship between their brain
0:12:56 and their body at this very elite level.
0:12:57 – Exactly.
0:12:59 Yeah, the other, I was gonna mention,
0:13:04 a lot of players and teams use this,
0:13:06 not just one time, but over time.
0:13:10 So they’ll get a scan, figure out a plan,
0:13:14 work on that for maybe three months or six months,
0:13:15 and then do another scan
0:13:18 and see how things are progressing and adjust accordingly.
0:13:21 So that’s definitely another way to,
0:13:23 in the long term, see if what they’re doing
0:13:27 is resulting in the change that they’re hoping to see.
0:13:29 – So what happened with that soccer player
0:13:33 who had the weak quadricep and other related troubles?
0:13:35 – Yeah, no, I think she’s doing great,
0:13:40 like staying healthy and getting ready.
0:13:42 – Yeah, so I know you can’t tell us her name,
0:13:44 but will we see her in the Olympics this summer?
0:13:45 – Yes.
0:13:46 – Great.
0:13:48 So as you were talking about that,
0:13:50 I mean, there was a moment where it was like,
0:13:52 okay, the athlete goes and gets the MRI,
0:13:56 and then you get the scan, you get the scan,
0:13:58 and then you said like, you crunch through the numbers,
0:13:59 and then you make the report.
0:14:02 Presumably, you crunching through the numbers
0:14:05 is like the result of many, many years of work
0:14:08 and kind of the core of what your company does.
0:14:10 So I wanna talk a little bit more about that
0:14:13 and kind of how you got here.
0:14:16 How did you come to start the company?
0:14:20 – How long do I have?
0:14:24 – A while, I mean, it’s not the radio, it’s a podcast.
0:14:26 – I’m warning you, I am a professor too,
0:14:27 so I can go on.
0:14:28 – Let me ask you this,
0:14:31 what was the moment when you decided to start the company?
0:14:36 – I can give you one moment, and then we can–
0:14:37 – Yeah, let’s try a couple of moments, sure.
0:14:39 – Yeah, let’s try a couple of moments in time.
0:14:44 – That’s what a good story is, like three moments in time.
0:14:46 – Right, right, three, not four.
0:14:49 – Four is a tricky number, we could do five or we could do one.
0:14:51 – Right, has to be odd, I think.
0:14:52 – Odd, I think, is better.
0:14:53 – Yeah, yeah.
0:14:55 So I’m a professor, I run a lab,
0:14:58 and for my entire career,
0:15:02 I’ve been fascinated with muscle and how it works.
0:15:06 And fascinated by something that we all,
0:15:10 in the muscle field, we call form function relationships.
0:15:13 So the idea that the way a muscle is shaped
0:15:16 and the way it’s structured and how big it is
0:15:20 influences how well it works, or how well it functions,
0:15:22 how strong it is, how well it behaves.
0:15:24 And there’s a lot to there, and there’s a lot of nuance,
0:15:27 and that’s like, I’ve spent a career studying that
0:15:29 in lots of different ways.
0:15:31 So I’ve always been interested in quantifying muscle
0:15:34 and figuring out how that influences how it works.
0:15:36 And both in healthy people or in athletes,
0:15:39 and also in different patient populations,
0:15:42 different, in particular, I have an interest
0:15:44 in movement disorders.
0:15:48 So neuromuscular diseases that lead to impairments
0:15:51 and mobility and movement ability.
0:15:55 So one of the light bulb moments for this
0:15:58 was the fact that I’d been using MRI
0:16:02 to study muscle in my research for a long time.
0:16:04 It’s kind of a ubiquitous tool,
0:16:07 or like often used tool in research.
0:16:11 But I was struck by the fact that
0:16:16 I was hearing from, in particular, a surgeon collaborator,
0:16:19 and the surgeon was telling me about his work
0:16:22 and helping kids with cerebral palsy
0:16:25 improve their movement, where they had hindered movement,
0:16:30 and largely because their muscles are impacted.
0:16:33 Not only do they have an impaired ability
0:16:36 to control their muscles, but their muscles end up
0:16:39 with impairments in their structure or their form,
0:16:41 which then influences how well they work.
0:16:43 So surgeons have to go in and do surgeries
0:16:45 to try to change that.
0:16:48 They do things like modify tendons
0:16:50 to try to make muscles less stiff,
0:16:55 or they transfer muscles to make them do a new thing.
0:16:58 But one of the big tricky parts is that oftentimes,
0:17:00 some of those muscles are very weak.
0:17:03 So if they choose the wrong muscle,
0:17:05 then they’ll make a weak muscle even weaker.
0:17:07 And that’s catastrophic.
0:17:10 So it’s a very fine line that a surgeon has to figure out,
0:17:11 and they have to go in.
0:17:14 We just talked about there’s 35 muscles in each leg.
0:17:16 So which of the muscles are the ones
0:17:20 that should be operated on, and which ones should be avoided?
0:17:25 And so what my collaborator, a guy named Dr. Mark Abel,
0:17:29 fantastic surgeon, he was telling me,
0:17:30 “Yeah, like it’s very hard.”
0:17:32 And I don’t, he didn’t have a way to see that.
0:17:35 All that he could do is look from the outside.
0:17:37 No technology could give him the information
0:17:39 he needed to figure out which muscles
0:17:41 he should focus on and which ones to avoid.
0:17:43 – Because it’s not obvious by looking,
0:17:46 what’s a strong muscle and what’s a weak one?
0:17:46 – Yeah.
0:17:49 – I guess that’s surprising to me on some level.
0:17:53 Like, I don’t, I’ve never thought about it,
0:17:56 but naively I would think you could look at a muscle
0:18:00 and say it looks strong or it looks weak, not so.
0:18:02 – Right, ’cause you just see it from the surface.
0:18:04 You don’t see it on the inside.
0:18:08 – And the other challenge is for every joint,
0:18:09 there’s many muscles.
0:18:12 So like we just said, the quadriceps has four muscles.
0:18:14 On the back of the leg, hamstrings,
0:18:17 there’s three hamstrings muscles.
0:18:19 There’s other muscles that are in the thigh.
0:18:22 So you’re just seeing what’s an impairment
0:18:24 in the overall movement, let’s say, of a joint,
0:18:27 but then there could be many muscles or combinations
0:18:28 of muscles that are leading to that.
0:18:31 And you don’t know when you look from the outside.
0:18:34 Our body is designed that way actually to be somewhat,
0:18:35 we call it redundant.
0:18:37 We have more muscles than we need probably,
0:18:40 but if you think about imbalances,
0:18:44 then any one of those muscles could create some trouble.
0:18:47 – So okay, so the surgeon describes this problem
0:18:50 he’s having, and then what do you do?
0:18:53 – So then I was thinking, well, you know,
0:18:56 that’s the information that we generate all the time
0:18:58 when we’re doing our research.
0:19:03 We take MRIs, we quantify, we identify muscles,
0:19:05 we create three-dimensional models of the muscles,
0:19:08 we figure out how they’re working from that.
0:19:11 But I was struck by the fact that none of that
0:19:13 was something that a clinician can use.
0:19:17 Despite the fact that MRI is obviously ubiquitous
0:19:19 in healthcare, right?
0:19:22 You can’t go to a hospital without finding multiple MRIs,
0:19:24 but there’s no way to use those MRIs
0:19:27 in the way that I was using them for my research.
0:19:30 And I thought, well, that’s too bad
0:19:33 because this would be very useful to the surgeon
0:19:37 in figuring out how to treat these patients.
0:19:40 So that was one light bulb at the beginning.
0:19:42 So a lot of it was figuring out
0:19:44 how to take something that we use in research
0:19:48 for these very specific targeted basic science questions
0:19:52 and turn it into something that is useful clinically.
0:19:54 – And again, this is like my ignorance.
0:19:56 Like I might have thought, well,
0:20:00 you could just do an MRI and see how big
0:20:03 or not big the muscles are and infer from how big
0:20:07 or not big the muscles are, how strong or not strong they are.
0:20:10 And that sounds straightforward,
0:20:13 but clearly it’s not.
0:20:15 Like why is it harder than that?
0:20:17 – So a couple of reasons.
0:20:20 One is going, taking the MRI pictures
0:20:22 and figuring out how big the muscles
0:20:25 is a very challenging problem.
0:20:28 So in order to accurately get how big the muscles are,
0:20:33 you have to essentially generate its shape in three dimensions.
0:20:35 So you have to get the whole length of the muscle.
0:20:39 And so you do that off of multiple MRI pictures.
0:20:43 So the MRI essentially kind of takes pictures
0:20:47 through the body at multiple different slices, we call them,
0:20:51 going from the abdomen all the way down to the feet,
0:20:54 sort of going cross-sectionally, we call it.
0:20:58 And so we usually have over 200 images like that.
0:21:01 So in each image, you have to find each muscle.
0:21:03 And so for any given image,
0:21:08 there’s probably at least 15 muscles or more.
0:21:11 – So it wasn’t like you could just push the like,
0:21:14 show me the muscles button on the MRI
0:21:15 and it would show you the muscles.
0:21:16 Like nobody had done that
0:21:18 and there was no obvious way to do it.
0:21:22 Certainly not for a surgeon ordering a standard MRI,
0:21:24 just didn’t exist.
0:21:25 – It did not exist.
0:21:30 – So you realize this, what happens?
0:21:32 How do you make it happen?
0:21:35 – So one of our first tasks was to figure out
0:21:37 how to get many muscles.
0:21:41 So one of the things that we had done on the research side
0:21:43 is really focus on a couple of muscles.
0:21:46 But I knew for this application, that wasn’t gonna work,
0:21:49 we have to be able to identify any muscle.
0:21:51 That was really the problem,
0:21:53 is that like you don’t know which one’s the problem.
0:21:54 So you don’t know which one to look at.
0:21:56 So you gotta look at all of them.
0:21:59 And then the next task was to figure out all those muscles
0:22:03 and figure out a process to go from the,
0:22:06 identifying each and every muscle and each and every image.
0:22:08 So it is called developing an atlas.
0:22:11 – And is that an AI problem?
0:22:15 – So now we have an AI and it’s the type of AI,
0:22:16 a supervised learning,
0:22:19 where you essentially train the computer
0:22:22 to do what the person would do.
0:22:23 But in order to do that,
0:22:25 you need to do what the person would do first.
0:22:31 And so we did that all manually at first.
0:22:34 In order to generate one of these reports,
0:22:39 at first it took us about 50 hours per person.
0:22:42 – Just going through image after image after image
0:22:45 and saying this is this muscle, that is that muscle?
0:22:46 – Exactly.
0:22:47 So we needed to develop that.
0:22:50 But the other piece we needed is this,
0:22:53 back to this normative database I talked about.
0:22:56 Because if I just told you how big your muscle
0:22:58 is in milliliters in volume,
0:23:00 what are you gonna do with that information?
0:23:01 Like, oh, great.
0:23:03 – And nobody knew, and it’s interesting,
0:23:05 it’s one of those things you always think,
0:23:06 oh, surely there’s some data in the world
0:23:07 that everybody knows X.
0:23:10 But so you’re saying nobody knew what was the kind of
0:23:13 median size of a particular quadricep
0:23:16 for whatever, a healthy 12 year old boy or whatever.
0:23:18 Nobody knew that at that time?
0:23:21 – Well, all of the information up until then,
0:23:25 for the most part, was based on dissecting cadavers.
0:23:30 Was based on taking cadavers and dissecting muscles,
0:23:33 weighing the muscles.
0:23:36 And one of the big challenges with that is usually
0:23:38 cadavers are older adults.
0:23:41 And so they’re not really representative
0:23:45 of a younger, healthy population.
0:23:47 And I will tell you at that time,
0:23:49 that that was a lot of work.
0:23:52 And I had people saying, like, why are you doing that?
0:23:55 Like, that seems like a waste of time.
0:23:56 That’s crazy.
0:24:01 I had this vision and I trusted that it was gonna
0:24:02 turn into something at least useful
0:24:04 to the research community.
0:24:06 And I’m thankful that we stuck with it.
0:24:10 – There’s lots more to come on the show,
0:24:12 including, but not limited to,
0:24:14 the work Sylvia and her colleagues
0:24:16 are doing with major league pitchers,
0:24:17 college football players,
0:24:20 and patients with degenerative muscle disease.
0:24:34 Sylvia and her colleagues trained an AI model
0:24:38 to do what had previously taken a human 50 hours
0:24:40 for every person who got scanned.
0:24:43 And they expanded from working with patients
0:24:47 with cerebral palsy to working with elite athletes.
0:24:50 Today, their clients include not just Olympic athletes,
0:24:54 but teams in the NBA and the Premier League.
0:24:56 Also, she told me they’re working on a project
0:24:58 with major league baseball.
0:25:02 – Yeah, so we’re working with the MLB studying pitchers
0:25:04 and we’re getting essentially a normative database
0:25:05 whole body scan of pitchers.
0:25:08 – And is that partly because like,
0:25:10 pitchers mess up their arms so badly?
0:25:12 Is that kind of the motivation there?
0:25:16 – Yes, there’s definitely a lot of issue
0:25:17 with injury and surgery.
0:25:21 And so the idea here is that by taking these scans,
0:25:25 we can really figure out where there might be weaknesses
0:25:30 and sort of potential areas for mitigating the injuries.
0:25:33 – So when you do work for a whole team,
0:25:37 like say the bowls, a basketball team, an NBA team,
0:25:40 like what’s the nature of that work?
0:25:42 What do you do for a team like that?
0:25:46 – Yeah, they will do a baseline of the whole team.
0:25:51 – And they basically tailor the athlete’s training,
0:25:54 presumably strength training in particular,
0:25:56 like on a muscle by muscle basis
0:25:59 based on the reports that you’re sending them.
0:26:00 – Correct, yeah, yeah.
0:26:04 – And I mean, you can imagine like better performance
0:26:06 being one outcome.
0:26:10 Reduced risk of injury seems plausible, right?
0:26:13 Like it seems obvious that like a big asymmetry
0:26:15 could make you more likely to be injured.
0:26:17 I mean, are you at a point now
0:26:20 where you can predict the risk of injury?
0:26:24 – That’s like a whole can of worms.
0:26:26 I won’t say that– – Let’s open that can of worms
0:26:26 for a sec. – Yes, let’s open it.
0:26:28 – I mean, is that interesting to you
0:26:30 or is that like too much or yeah?
0:26:33 – No, no, no, this is something we think about a lot.
0:26:38 And let me, I wanna, so first I’ll tell you
0:26:39 why it’s a can of worms.
0:26:42 But then I’ll tell you what project that we’re doing.
0:26:43 – Then tell me about the can.
0:26:44 We’ll look at it from the outside.
0:26:46 – So from the can of, like there’s a lot of technologies
0:26:50 out there that will say that they’re predicting injury risk.
0:26:51 They’ll give you numbers
0:26:54 and they’re just not based on anything.
0:26:59 And so I don’t know, there’s a lot of–
0:27:02 – So it’s the land of specious claims, yeah.
0:27:03 – Yeah, yeah, yeah.
0:27:05 So that’s not what we’re about.
0:27:09 We’re about like providing actual things that matter.
0:27:14 And so the question is like, can you do these muscle scans?
0:27:20 Do they correlate with injury likelihood in some way?
0:27:23 And so we actually have a project
0:27:24 to address that very question.
0:27:26 It’s actually funded by the NFL.
0:27:29 We’re actually in that project.
0:27:32 We’re working with college teams.
0:27:33 – College football, college football.
0:27:34 – College football teams, yeah.
0:27:39 Baselining entire rosters at the beginning of the season.
0:27:42 And then tracking hamstring injuries.
0:27:45 And then if an athlete gets injured,
0:27:47 they come back for a scan at the time of injury
0:27:50 and then at return to sport.
0:27:54 And so one of our questions is based on the baseline scan,
0:27:58 can we predict who’s more likely to get an injury,
0:28:00 an initial injury, index injury?
0:28:02 And then the secondary question is,
0:28:05 can we predict who will be re-injured?
0:28:08 Saying that we often people pair it with other things.
0:28:11 In this project, we’re always also doing that.
0:28:13 Each athlete is getting an assessment
0:28:15 of their sprint mechanics.
0:28:18 So kind of the biomechanics of how they run.
0:28:22 And then also assessment of their strength,
0:28:23 of their hamstring muscle.
0:28:25 So kind of measured strength.
0:28:27 Obviously you can’t do that when they have an injury,
0:28:30 of course, but you can when they’re healthy.
0:28:33 – That biomechanics piece seems like something
0:28:36 that has been developing in parallel with your work.
0:28:39 Also driven by computer vision, right?
0:28:44 The like markerless motion capture seems like a big world
0:28:49 that overlaps with your world some.
0:28:54 So tell me about your work with female athletes
0:28:59 versus male athletes and how that plays a role.
0:29:01 – Yeah, I will say probably the biggest thing
0:29:05 that we’ve been focused on is making sure
0:29:07 that our data addresses that.
0:29:12 So our normative database is separated by sex.
0:29:18 And it is different because women aren’t small men, right?
0:29:22 So it’s important that we have that basis to compare
0:29:27 that’s like for women and not comparing to some average
0:29:29 or primarily male data set.
0:29:31 So that’s one huge important thing
0:29:35 is that it’s compared to the normative values
0:29:37 for the female population.
0:29:42 And then in terms of like working with the female athletes,
0:29:46 I think one of the big ones is really just ability
0:29:50 to personalize and provide this
0:29:55 like really accurate detailed assessment of their bodies.
0:29:58 And a lot of the knowledge
0:30:02 about like appropriate body composition
0:30:04 historically has been based on studies in men.
0:30:08 And but then we’re applying them to women
0:30:11 and making us feel really bad about ourselves.
0:30:15 So really motivated to move away from that
0:30:18 and sort of acknowledge the muscular physiology
0:30:20 and anatomy of the female
0:30:25 and also the female athlete to really understand that.
0:30:28 I think one thing obviously that we’ve seen
0:30:32 is ACL injuries are more common in women than men
0:30:37 and examining how these like recovery profiles look
0:30:40 and how they differ between men and women.
0:30:41 That’s something that we’re observing
0:30:43 and seeing how those things shake out.
0:30:47 But we’re motivated by really providing that information
0:30:48 that’s specific to women.
0:30:53 – So what are some of the non sports things
0:30:55 you’re working on, things you’re trying to figure out?
0:31:00 – Yeah, I mean, one that I’m really interested in
0:31:03 is this area that we’re applying to
0:31:06 in clinical trials for muscle disease.
0:31:10 So we’ve been working in a specific muscle disease
0:31:15 called fascioscabular humeral muscular dystrophy, FSHD,
0:31:19 which is a slowly progressing muscle disease,
0:31:20 genetic in basis.
0:31:25 And so eventually people with FSHD need a wheelchair
0:31:28 just life is very difficult.
0:31:30 And so it’s pretty devastating.
0:31:31 But the other exciting thing
0:31:33 is there are some new treatments out there,
0:31:36 some in particular gene therapies coming online.
0:31:39 And now the challenge is, do they work?
0:31:42 ‘Cause the problem is in these diseases
0:31:44 ’cause they’re pretty slowly progressing.
0:31:49 If you wanna see if a drug is helping somebody,
0:31:52 it’s very hard to see that in a slowly progressing disease.
0:31:55 – The clinical manifestations are hard to pick up.
0:31:57 – They are. – If it makes your muscles
0:32:01 shrink more slowly, it’s gonna be hard to see.
0:32:02 – It’s very hard to see,
0:32:04 especially from like rudimentary measures.
0:32:07 But with the MRIs, we’ve been able to provide
0:32:09 this really detailed insight
0:32:11 about the disease state of each muscle
0:32:14 and how it’s progressing over time.
0:32:17 And so one of our goals is to really lean in on this
0:32:21 and help figure out exactly how people should look
0:32:23 at all this data and figure out
0:32:25 if a drug is working or not.
0:32:28 It’s really profoundly important because without that,
0:32:30 these clinical trials just won’t move forward.
0:32:37 – What else are you sort of still trying to figure out?
0:32:40 – So we talked about predicting injury,
0:32:44 but having all the data needed to show,
0:32:47 like if your scan looks like this and if you do this,
0:32:50 you will be able to improve your jump height by that.
0:32:53 – Yeah, you’ll be able to throw a fastball
0:32:54 two miles an hour faster.
0:32:57 Like that would be wildly valuable.
0:33:00 – That would be very, and we do have data in our research.
0:33:05 We were able to show that these muscle scores
0:33:07 correlate with performance metrics,
0:33:09 such as jump height and speed.
0:33:11 So we for sure see that.
0:33:13 The question is then the spin on like observing
0:33:15 how that plays out.
0:33:19 Like if you then strengthen the appropriate muscles,
0:33:21 how much faster do you get?
0:33:23 And you just need more and more data
0:33:26 to really like to go after that.
0:33:28 But that’s one thing that I’m fascinated by.
0:33:30 One of the other interesting ones,
0:33:32 I’ll go, can I go off on a tangent?
0:33:33 – Any, anything you want.
0:33:36 – One of our research partners
0:33:39 that’s interested in how muscles adapt to strength training
0:33:42 and different interventions and what influences that
0:33:45 had a really interesting finding
0:33:48 that I think is quite profound, but also obvious.
0:33:51 So everybody, if they’re targeted training,
0:33:53 their quadriceps and hamstrings,
0:33:56 those muscles got bigger.
0:33:57 That makes sense.
0:33:59 But in a fair number of the people,
0:34:01 some other muscles got smaller.
0:34:06 And then he had done some controlling
0:34:10 and documentation of nutrition intake.
0:34:14 And he found that people that had higher caloric
0:34:19 and protein intake had less of that effect.
0:34:21 – So all the gym bros telling you
0:34:25 to eat a lot of protein are validated by this guy’s study.
0:34:28 – Yeah, but it’s not necessarily to make that muscle
0:34:29 that you’re working bigger.
0:34:30 So you don’t lose the other muscles.
0:34:34 – That’s a good one.
0:34:37 And was he using your scans to figure that out?
0:34:38 – Yeah, he was using our scans.
0:34:41 And the thing that was cool is that normally
0:34:43 in research, you wouldn’t bother looking
0:34:44 at those other muscles.
0:34:46 You would just look at the ones that were targeted.
0:34:48 ‘Cause those are the ones you just think about.
0:34:52 But by getting the entire extent of all the muscles,
0:34:55 you see these impacts that you wouldn’t necessarily
0:34:56 have noticed.
0:34:58 – Like he wasn’t even looking for it.
0:34:59 – Yeah, it’s quite profound
0:35:01 because somebody’s strength training,
0:35:03 recovering from an injury,
0:35:08 that really means the nutritional element’s important
0:35:10 because you could be strengthening some muscles,
0:35:14 but weakening others if you’re not playing your cards
0:35:15 right there.
0:35:20 – We’ll be back in a minute with the lightning round.
0:35:33 I wanna finish with the lightning round.
0:35:34 Won’t take too long.
0:35:35 It’ll be fun.
0:35:36 – Okay.
0:35:37 – What?
0:35:38 – I don’t know what that is.
0:35:40 – Well, you’ll find out right now.
0:35:43 – Does this have to be fast?
0:35:43 – Nope.
0:35:44 (laughing)
0:35:46 – I could call it the random round.
0:35:48 – Okay, I like that, random.
0:35:52 – Have you scanned yourself?
0:35:53 – Yes.
0:35:54 Multiple times.
0:35:55 – Of course.
0:35:56 – Of course.
0:35:58 – I guess what kind of thing are you talking about?
0:35:59 I’ve probably been in an MRI machine,
0:36:01 I don’t know, maybe a hundred times,
0:36:02 like a lot of times.
0:36:03 – It’s not radiation, right?
0:36:04 It’s not like an x-ray.
0:36:05 You could do it every day if you want to.
0:36:07 – Yeah, as much as you want, yeah.
0:36:08 – What’d you learn?
0:36:11 – So I actually used it,
0:36:13 I’ve learned lots over the years,
0:36:15 but I will tell you one anecdote,
0:36:17 I have a hip replacement.
0:36:23 I have a genetic condition that leads to early arthritis.
0:36:27 And so I was, before I got my hip replacement,
0:36:28 I got a scan.
0:36:30 I knew I was getting weak,
0:36:33 but holy cow, was I really weak on that side?
0:36:38 What was profound was how weak my hip flexors were,
0:36:40 very weak.
0:36:42 And I think a lot of times people talk
0:36:44 about hip flexors being tight.
0:36:47 And that’s kind of what I thought was happening.
0:36:51 I felt pain and I felt like I was having a lot of tightness,
0:36:53 but it was actually weakness.
0:36:57 And they were like super small on both sides,
0:37:00 but really especially on the side that was affected.
0:37:04 So that was one thing that I worked on a lot.
0:37:06 – Did that genetic condition you have,
0:37:08 did that influence your work at all,
0:37:10 your decision to go into the field?
0:37:15 – I mean, like loosely maybe,
0:37:18 because my dad had the same thing,
0:37:20 which actually caused him to go blind.
0:37:21 – Oh goodness.
0:37:23 – It has like a multiple different issues.
0:37:26 And so I think that at a early age got me interested
0:37:31 in medicine and disabilities and helping people.
0:37:37 So that might be broadly speaking.
0:37:38 And I knew I had some eye problems.
0:37:40 I didn’t know the genetic thing.
0:37:42 We didn’t discover that till later, but.
0:37:43 – Huh, interesting.
0:37:48 What’s the most underrated muscle in the human body?
0:37:52 – That’s a hard one.
0:37:57 So I have a few favorite muscles.
0:37:58 – Okay, what’s your favorite muscle?
0:38:03 – Yeah, so the psoas muscle, psoas major, it’s a hip flexor.
0:38:05 But it also, it’s really cool.
0:38:09 It actually, it’s also a back muscle, lower back muscle.
0:38:11 So it attaches to the lumbar vertebrae,
0:38:15 but then it also crosses the front of your hip.
0:38:16 It’s really hard to find
0:38:19 because it’s like really back deep in your hip.
0:38:21 It goes right over your, your femur.
0:38:24 – So it goes like in the middle of your body kind of.
0:38:25 – Yeah, right in the middle.
0:38:26 It kind of connects everything.
0:38:29 Sort of connects your lower extremity
0:38:31 to the rest of your body in some ways.
0:38:34 – Okay, last one.
0:38:38 – Why do you hate astrophysicist Barbie?
0:38:41 – I don’t hate anything.
0:38:42 – I mean.
0:38:45 – Well, it’s too perfect.
0:38:50 It’s kind of like this, you know, idea that like,
0:38:55 oh, you know, you can inspire girls to go into science
0:38:56 by showing them that Barbie does too.
0:38:59 But Barbie is like fictitious.
0:39:00 So it kind of tells you that
0:39:06 like promotes the idea of perfectionism in society,
0:39:09 but definitely in girls.
0:39:11 And, you know, what we really want to promote
0:39:14 is almost the opposite of that, is taking risks
0:39:18 and not worrying about being perfect
0:39:21 and just doing something that matters to you.
0:39:23 So yeah, I don’t know.
0:39:24 I mean, I don’t hate it.
0:39:26 I had Barbies when I was a kid, but I just,
0:39:28 it kind of like, so it’s something that-
0:39:29 – You wrote a whole column.
0:39:32 You wrote a whole column in the newspaper about-
0:39:33 – I did.
0:39:34 – About why it all worked out.
0:39:35 – Yeah, wasn’t that cool?
0:39:36 I was very proud of that.
0:39:39 Yeah, no, and really that, you know,
0:39:43 the astrophysicist part honestly was more of a hook.
0:39:45 The article I had written already
0:39:48 before that astrophysicist Barbie came to be,
0:39:51 it was about the issue of perfectionism
0:39:56 and how that dissuades girls to go into STEM and research.
0:40:02 – Well, a good hook is important, and you found one.
0:40:03 – Yeah, it was perfect, yeah.
0:40:10 – Sylvia Blemker is a professor at the University of Virginia
0:40:14 and the co-founder of Springbok Analytics.
0:40:16 Next week on What’s Your Problem,
0:40:17 I’ll be talking to Jimmy Buffy.
0:40:21 He is using AI to bring the insights of biomechanics
0:40:24 to professional athletes.
0:40:26 Jimmy told me that before the advent of AI,
0:40:29 when biomechanics experts tried to work with athletes,
0:40:32 it could be somewhat awkward.
0:40:34 – So you’ve got like a picture in his underwear
0:40:36 with a bunch of little metal balls.
0:40:38 They’re like, just pitch like you always pitch.
0:40:40 – Right, and the state of the art for tracking it
0:40:42 was an awful experience
0:40:45 for the people you were trying to track.
0:40:46 – So what changes?
0:40:49 – The big inflection point was computer vision,
0:40:51 basically using artificial intelligence
0:40:55 to identify where those joints are in a camera image
0:41:00 rather than needing to paste those reflective markers.
0:41:03 – Today’s show was produced by Gabriel Hunter-Chang,
0:41:05 edited by Lydia Jean-Cott
0:41:08 and engineered by Sara Bouguere.
0:41:12 You can email us at problem@pushkin.fm.
0:41:14 I’m Jacob Goldstein, we’ll be back next week
0:41:16 with another episode of What’s Your Problem.
0:41:19 (upbeat music)
0:41:21 (upbeat music)
0:41:25 (upbeat music)
0:41:27 (upbeat music)
0:41:37 [BLANK_AUDIO]
On the next few episodes of What’s Your Problem, Jacob Goldstein is talking with people working at the frontiers of technology to help elite athletes perform better.
Today’s guest is Silvia Blemker, a professor of biomedical engineering at the University of Virginia and the co-founder of Springbok Analytics.
Silvia’s problem is this: How do you combine MRI scans and artificial intelligence to generate new insights that can help both elite athletes and people suffering from diseases that affect the muscles.
Springbok’s sports clients include medical researchers, Olympic athletes, Major League Baseball and several professional basketball and soccer teams.
This summer, a bunch of Pushkin podcasts are coming out with Olympics-inspired shows. Revisionist History has a series about America’s decision to participate in Hitler’s Berlin Olympics in 1936. The Happiness Lab has an interview with a coach who coaches coaches. And Cautionary Tales tells the story of the family feud that gave us both Puma and Adidas.
See omnystudio.com/listener for privacy information.