Sherlock Holmes Was Right – We Need to Become Movement Detectives.
If you haven’t already, please read Part I.
I have quoted Claude Benard when I give presentations on why we need to develop better assessments to test the neuromuscular capacities of athletes after ACL injury. The quote is:
“[Those] who have excessive faith in their theories or ideas are not only ill-prepared for making discoveries; they also make poor observations.”
I use this quote when speaking about the mistakes that often get made when an athlete is “cleared“ to get back to sport too early. We all want to focus on the positive, and because bias is a part of the human condition, we tend to overlook the negative. It‘s at this point that mistakes get made and we miss the deficits the athlete is hiding, even though they are sitting right in front of our eyes. In short, we have too much faith in our ideas and beliefs, we tend to think our practitioner‘s eye is accurate enough for decision making, and we roll the dice.
Figure 1: Experience and clinical judgement are key. But we need data to support the practitioner‘s eye.
We also might avoid a good post-mortem or a debrief to determine what we could have done differently in the event of a reinjury. Instead we chalk it up to bad luck.
Let‘s take a second to ideate grandly with a look into the future.
The future with Norman the AI is that it will most likely learn to extract insights from our movement patterns to mitigate some of the pitfalls around return to sport decision making. That‘s the cool thing about machine learning; it works in a very similar way to how the human brain learns. Just how experienced practitioners probably do have an eye for picking out errant movement strategies, Norman will likely be able to extract the nuance from our movement patterns that can help find athletes who need a bit more time to train before returning to sport.
Just a heads up – Norman can already predict when someone had an ACL injury to within a few weeks. The more data it is fed, the better the predictions. This opens up huge potential for us because Norman can tell us whether an athlete is tracking according to expectations, behind expectations, or ahead of expectations, and it can do this based on objectively determined information that is recorded from our everyday movements.
This will no doubt be a huge revolution in the world of physical medicine and rehabiliation.
But Norman still has some work to do.
As we discussed in the first blog post, the challenge with determining when an athlete is ready to return to sport after injury is that we are dealing with complex-adaptive systems. Complex-adpative systems can‘t be understood with a few functional tests, especially when those functional tests are performed nowhere near the inciting event for the reinjury. Remember the example about predicting a tornado next summer from today‘s weather data? This would probably provide low predictive validity.
Nevertheless, we still need to manage athletes returning from injury more effectively. And we can use heuristics to help manage the complexity and uncertainty that arises with complex-adaptive systems.
Here are a few heuristics that I like to follow when it comes to managing decision making for athletes as they are returning from injury:
- It‘s an n=1 game so treat an athlete‘s journey back to health, sport, and performance on an individual-by-individual basis (i.e., steer away from the generalized timeline approach).
- Instead, use a functional milestone roadmap in advance that will support decision making – this is a “green means go“, “yellow means proceed with caution“, and “red means stay where you are and keep training“ approach to guide the rehabilitation process. Ideally, these milestones should be measurable.
- If the athlete was strong to begin with, use pre-injury baseline data as a benchmark to measure progress.
- If this data doesn‘t exist or if the athlete was not strong to begin with, use benchmark data from a comparative group to measure progress.
- If we don‘t have adequate data from a comparative non-injured group, we can use the non-injured limb as a benchmark to measure progress – this is called a limb symmetry index or a limb asymmetry index. This is ok, but we need to remember an injury on one side often impacts the other side – we can find athletes who are symmetrical but weak on both sides.
- Remember that the process of return to health, return to sport, and return to performance after injury is non-linear – athletes may experience progressions and regressions. This means we can‘t treat the return to sport readiness test like a final exam where the athlete crams, passes the test, and then resumes their activities without any follow up or monitoring. Instead, we need to treat this whole process like a transitional developmental pathway and we need to monitor their function and performance.
- Finally, we need to be like Sherlock Holmes: search for clues that can be objectively verified and start each evaluation timepoint or meeting about an athlete‘s rehabilitation plan with these clues. Lead with the evidence!
Figure 2: An example functional milestone roadmap tailored to each person (n=1) using symmetry, strength, and performance metrics with bookended baseline profiles (use benchmark data if there’s no initial baseline).
“It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.”Arthur Conan Doyle, A Scandal in Bohemia
Figure 3: Treat each injury and return to sport project like Sherlock Holmes. Search for clues.
To manage the uncertainty surrounding complex-adaptive systems, like athletes returning from injury, we need to be objective and data can with this. I’m not trying to imply that clinical judgment and experience aren’t valuable here; they are key as well.
Figure 4: We are constantly balancing our experience and judgment with the facts. A good rule is to be nurturing our instincts but constantly searching for facts. But when we find facts that don’t fit with our beliefs, we need to change our beliefs and not twist the facts.
But, if left unchecked, subjectivity may lead us to twist facts and distort reality. Just like the quote says, starting a conversation about an athlete’s return to sport readiness without data may lead to us to commit a capital mistake.
Figure 5: We are all prone to confirmation bias. It’s a natural part of the human condition. We need to account for this eventuality using data to avoid the regrettable cases of athletes who sustain reinjuries due to physical limitations that were trainable and missed.
We can be tricked into thinking that an athlete is sufficiently prepared to move onto the next functional milestone or return to sport because we all can find creative ways of altering our movement strategies to compensate for the real deficits. At worst this could make us more susceptible to reinjury, or at best limit sport performance.
Figure 6: Athletes are complex-adaptive systems and injury prevention is tough due to the complexity and the uncertainty. Rather than getting hung up on preventing an injury, let’s focus on preparing our athletes physically and mentally for the demands of sport. We will enhance performance and we might even prevent a few injuries along the way.
But what type of data do we need and how can we use data to improve our decision-making process? This is a huge challenge because many of us in the world of sport performance or rehabilitation don’t actually use objective data to make decisions.
Again, let’s turn to Sherlock Holmes.
Think about your favourite Sherlock Holmes mystery. He showed up to the scene of the accident with his magnifying glass, fingerprint kit, and specimen jar to look for clues so that he could arrive at a plausible explanation for “who killed the butcher”.
This is sometimes called a Bayesian approach or inductive reasoning. The same process can be applied for monitoring and assessing an athlete across a functional milestone continuum as they recover from injury. This is a great way to begin to wrangle the complexities of managing the return to sport process after injury.
Figure 7: Deductive and inductive reasoning are two ways humans can arrive at understanding the natural world. Inductive reasoning allows us to go from a set of isolated experiences to building testable hypotheses. We can use inductive reasoning in return to sport to improve return to sport management by uncovering the physical limitations that are holding the athlete back.
The only difference is that unlike Sherlock Holmes, our measurement instruments need to target important capacities like maximal muscle strength, rate of force development (RFD), and power. They also need to target functional abilities like walking, running, jumping, change of direction movements, and, most importantly, movement ability on the field of play in a sport-specific manner. We need to use a battery of tests to find clues and we need our testing methods to be versatile.
I often say this to friends and colleagues who contact me for advice on using dual force plate systems to capture force-time asymmetries in jumping – it’s a great tool, but don’t hang your hat on one or two tests. Instead, build an expansive and sport-specific test battery and build a functional milestone roadmap.
This is where Plantiga addresses a very important gap. We can measure capacities like lower body maximal muscle power, running speed, walking speed, and vertical jump height. We can measure movement abilities like stride length, stride variability, ground contact time, and limb speed in walking and running. And we can take the system onto the field, court, snow, or ice to measure movements in a real-world environment, where they likely matter most. We can also move from a single discrete timepoint of testing to a ubiquitous monitoring system that flags potential problems before they have disastrous consequences, like a full-blown reinjury.
Dr Matt Jordan PhD
My name is Matt Jordan. My PhD is in Medical Science. I’m an applied sport scientist working with elite athletes. Head to my website: www.jordanstrength.com