My Thoughts on the 2019 MIT Sloan Sports Analytics Conference Injury Panels

By Dr. Ankur Verma


MIT Sloan Sports Analytics 2019 injury panel

As I have done every year since 2013, I made my way to Boston in late February/early March for the MIT Sloan Sports Analytics Conference.  Over time, given my background, I’ve increasingly paid attention to the health/injury panels at the conference.  While I am interested in all aspects of sports analytics, including player performance, drafting, and the like, I have found myself dabbling in the injury side of things as, to this point, it matches my knowledge and training.

The major injury panel this year was called “The Performance/Precaution Tradeoff: Player Health”.  I was intrigued by the name of the panel as it was suggestive of some of the hot topics that dominate injury conversation today.  At what point should a player be rested to prevent him/her from entering a “danger zone” of increased risk of injury?  When is it better to play a back-up for an injured player? 70% health?  60% health?  And how does some of the new data being collected infringe upon privacy rights?

These were just some of the things that popped in my head.  Did the panel play out the way I expected or did it bring up other topics I hadn’t considered?

I took some rough handwritten notes in real-time.  One takeaway: my handwriting has gotten increasingly worse with time.  I guess that’s the byproduct of being a doctor.  I plan to re-watch the panels once the videos are released, but because they aren’t available right now, there may be some discrepancies between what I wrote and what was actually said.  So apologizes in advance if there’s some misinterpretation or misrepresentation in what follows.

The following is the general flow of (paraphrased) conversation in this panel based on my notes (or at least what was legible enough for me to read).  It’s not inclusive as it is based only on the notes I took, as there are some topics I missed or didn’t write down.  I’ll occasionally chime in with my thoughts in bold.

A caveat, and running joke, is that the panelists at the MIT SSAC often use general phrasing to avoid giving away proprietary secrets.  This is no different in some cases of this injury panel.

The panelists consisted of Casey Smith (Head Athletic Trainer, Dallas Mavericks), John DiFiori (Director of Sports Medicine of NBA, Hospital for Special Surgery), and Adir Shiffman (Executive Chairman of Catapult).  The panel was moderated by Gretchen Reynolds (New York Times “Phys Ed” columnist).

  • Evaluating connective tissue and joint structure has been an important area of injury prevention over the past five years
  • An example of this is tendinopathy (essentially injury to the tendon)
  • Now we can measure “changes in the tendon” (it is unclear what changes Smith exactly meant)
  • Due to some of these measurements, there are things that we previously thought we had to manage that we now know that we can improve
  • The last point is intriguing.  Soft tissue injuries (basically muscle, tendon, and ligament injuries…the “sprains”, “strains’, and “contusions” of the world) comprise a large part of athlete injuries.  They’re nagging, affect performance, and can evolve into bigger injuries.  By “managing” some of the factors that may lead to soft tissue injury (in this case, tendinopathy), we are doing our best to extend an athlete’s career.  But if some of that management has been shifted to “improvement”, as Smith implies, there’s a chance that not only are we extending an athlete’s career, but there is potential for better performance as an athlete ages.  Perhaps a player’s “prime window” is extended (of course, other factors are involved).  The “nicks” and “dents” that an athlete accumulates as he plays into his career can potentially be improved earlier in his career…we are seeing the most optimal form of the athlete.
  • Loading and ankle instability can affect tendinopathy
  • It is important to “individualize” when it comes to injuries
  • Do not overreact to blips (use Big Data)
  • Individualization is obviously important when it comes to injuries.  On the other hand, while there is a trap to overly generalize when it comes to injuries, I also think using models or data to determine trends or general expectations when it comes to injuries is a useful tool when it comes to decision-making.  I think that is echoed in DiFiori’s third bullet point.  While individualization is important, you have to be able to recognize an outlier, or at least wait for more data before determining whether a trend is true.  Big Data is also a common theme when it comes to DiFiori during this panel, something that was great to hear from a sports medicine physician.
  • Applying this information is still a challenge
  • Give the player context
  • Gives an example of a player who says he feels great, which contradicts the data
  • You can have all the data you want, but if you can’t implement it, it’s useless.  Implementation is a direct result of buy-in from those in the trenches: the players, coaches, and even management.  The key is helping these parties understand what the data is, why you’re collecting it, and how it can improve long-term health and decision-making.  An underrated notion, based on my experience, that gets lost in all the wearable tech hubbub is that players will be interested if the data is something that can help them improve their careers–whether through playing or injury prevention/management.  In the heat of the moment, I’ve found it’s obviously hard for coaches and players to be unemotional (nor would I want them to be) as they’re trying to win and competing is in their genes.  But this is where the value of having an impeccable communicator (and enabling the communicator to influence decisions) is a factor.  This helps put everyone–from executes, coaches, and players–on the same page.  Not by groupthink, but by promoting understanding throughout the organization.
  • NBA wearable tech committee
  • Committee is measuring accuracy of wearables
  • Also looking at injury prevention by using data that is measuring things differently or more precisely than the past
  • For example, measuring the actual load, as opposed to minutes played (which was previously used a surrogate for load)
  • Reminds me of some NBA injury analysis I did about six years ago, using minutes played as a surrogate for load.  Unfortunately, using surrogates in this manner may be necessary in the arena of grassroots injury analytics as a lot of better data is understandably proprietary and, in the instance of health, related to privacy.  It is therefore difficult to get access to.
  • Discusses modifiable injury factors
  • You need someone who understands the data and communicate it to the medical staff and players
  • You need to have a professional as a part of that group, not just a coach who crunches numbers
  • This speaks to my point above and is absolutely critical.  Based on the criteria DiFiori has set forth thus far, the ideal candidate would be a data scientist with a sports medicine background who is a superior communicator.
  • How do you integrate disparate information, such as heart rate with blood testing?
  • By combining injury data with game load data
  • If I remember correctly, this answer was vague.  It obviously would be nice to know specifics, but it’s understandable why such a thing wouldn’t be divulged in this setting.
  • It is crucial to replicate the load of games in practice
  • For example, cheerleaders running backwards
  • This concept is crucial and, unfortunately in my experience, something that needs to be picked up by more organizations.  There needs to be more critical thinking done about the drills used in practice.  Risk/benefit analysis needs to be applied.  Is this drill functional?  Regardless of the answer, what evidence do we have to back it up?  What is the risk of this drill?  Is there far more downside than upside?  What are we trying to accomplish by this drill and is that accomplishment translatable to games and something worth pursuing it?
  • Mental health is an untapped area of research
  • You can measure “heart”
  • For instance, hustling after a loose ball shows “heart”
  • This would be a high load under anaerobic conditions
  • Single sport specialization in youth
  • Rest and recovery essential for making progress
  • Important to emphasize this to parents, since they are usually “go, go, go”
  • This “go, go, go” attitude lends itself to overuse injuries, such as tendinopathy
  • Best advice to youth: “don’t play when you’re hurt”
  • Sleep, diet, and recovery
  • Organizational culture comes into play for these factors
  • Food diary and logs are up to the player
  • Everything is about decreasing exposure to risk to injury
  • Virtually all the companies (presumably wearable tech and the like) that present to him have internal research
  • That means they are not peer-reviewed
  • Agrees.  Asking for research wipes out 90% of what comes through the door
  • Return-to-play decision-making has evolved
  • In the past, they would measure speed and agility to see if a player was ready
  • Now they use musculoskeletal ultrasound and force plates and compare the results to a preseason baseline to guide RTP decision-making
  • Changing the mindset of coaches: When you’re asked if a player is at risk for re-injury, the answer is always yes.  But need to quantify
  • Ultrasound is one of the techniques in medicine that is rapidly advancing.  It is quick and easy to do at bedside and is done in a few minutes, a massive time- and expense-saver compared to an MRI.  And again, quantifying the actual risk helps coaches and decision-makers visualize this concept when you present to them.  Of course, if a coach has made up his mind one way or the other, then it doesn’t matter what you do.  That’s why it’s important to have organizational buy-in from top to bottom and have the right people empowered so that what they say carries weight.
  • An example of something we don’t have data for: if one hamstring is 10% within the other in terms of strength, what is the risk of injury?
  • Was asked to give an example of an interesting finding in data analysis
  • Cites the way you practice
  • For instance, sprints in the half court result in increased load due to acceleration/deceleration in a small space
  • Agrees.  For example, in cricket, bowlers were practicing fastballs at 60-70% of max speed
  • Then there was an increase in injuries on game day because they were suddenly bowling at max speed
  • Gives an example of rugby.  My notes are a little unclear, but it seems he insinuates that increasing contact during rugby practice may decrease injury risk in games.
  • I’m not sure about that last point and whether I agree with it or not, but the main takeaway here is that practice comes up again.  For Shiffman, replicating game conditions in practice seems to be a common theme in reducing injury risk.  Whether you agree with that or not, I think it’s important for organizations to review practice strategies.
  • Equates this to running and how some runners never train as hard as the actual race
  • From my extensive race coverage experience, I personally can attest that this leads to injuries.


So that’s the gist of the panel.  I think the greatest takeaways are these: individualization, communication, and practice conditions.  Organizations would do well to review their methods in each these realms and determine areas for improvement.

The other interesting injury panel I attended was “The True Cost of Tommy John Surgery in Major League Baseball”, which was given by Dr. Eric Makhni, an orthopedic surgeon with Henry Ford hospital in Detroit and a current team physician for the Detroit Lions.  As far as youth sports, Makhni echoed a common theme in that increased specialization means you are not giving your child enough time to rest.  It’s time to listen, parents!  The rest and recovery period is just not there.

Table of results for the Tommy John algorithm presented by Dr. Eric Makhni.  When is it better to do surgery vs. non-operative treatment?

The other interesting thing is that Makhni revealed an algorithm that is being used to determine whether a player needs to undergo surgery for partial UCL tears, as opposed to non-operative treatment such as rehab and regenerative medicine (which in itself is a topic for another post, particularly how it relates to UCLs).  The algorithm incorporates factors such as timing of injury (for instance, spring training vs. playoffs as it would alter the timeline of recovery), age, WAR, etc. to project the likelihood a pitcher will return to meaningful play (based on projected WAR).  While some can quibble with the methodology (WAR is a cumulative statistic, so how does this affect starters vs. relievers?), I believe it’s a huge step in the right direction in the intersection of analytics and sports medicine.

The Malcolm Gladwell-David Epstein one-on-one was as fascinating as you might expect, and they even touched on injuries briefly.  One of them (I didn’t write down who it was) said that early (and late) diversity in terms of “physical skills” could be a protector against overuse injuries.


Well, that’s what I have.  Once the videos are released, it’s possible I’ll have some more thoughts that I’ll post.

Leave a Reply

Your email address will not be published. Required fields are marked *