Here are three studies the Driveline R&D Department read this month:
The Relationship between the Glenohumeral Joint Internal Rotation Deficit and the Trunk Compensation Movement in Baseball Pitchers
One of the more commonly monitored physiological effects of overhand throwing is the deficit of throwing shoulder internal rotation.
This is known as glenohumeral internal rotation deficient (GIRD) and is defined by a decrease in pitchers’ maximum internal rotation of the shoulder joint. Most of the time when this is discussed, it’s from the perspective of how to regain a “normal” range of motion in the athletes’ throwing shoulder.
Cheng et al. in Taiwan compared an experimental group of pitchers who had GIRD to a control group of pitchers who did not have GIRD with a VICON motion capture system. They compared 18 different kinematics, including torso, pelvis, and hip positions. The authors concluded that the athletes with GIRD had a lower total range of motion of the drive hip and were more rotated with their torso at ball release.
Descriptive studies like these don’t necessarily tell us if there are causal relationships. For example, we don’t necessarily know that more torso rotation at ball release is an effect of having GIRD or if it is from some other difference between the group, but it allows us to use some theory to think about further testing.
It makes some theoretical sense that athletes who have a reduced capacity to internally rotate their throwing shoulder could compensate by rotating their torso more to achieve a similar outcome of the pitch, but that is more or less speculation at this point.
More studies would have to be done to be sure of anything, but this approach to understanding compensatory effects of GIRD is interesting, nonetheless. Instead of using biomechanics as a tool to change athlete’s movements closer to a single movement profile, this idea could be used to have them find a solution best for their individual movement constraints.
Group analysis of biomechanics gets a bad rep since technique is such an individual problem to solve. Many athletes have unique training and movement constraints, but creative group analyses like these will get the industry closer to where we can provide individual training prescriptions with motion capture assessments like the ones used at Driveline.
Comparative Pitching Biomechanics Among Adolescent Baseball Athletes: Are There Fundamental Differences Between Pitchers and Non-pitchers?
Biomechanical comparisons between how pitchers and position players throw are rare, if not non-existent. Dr. Bullock et al. at Wake Forest took a look at how pitchers’ and non-pitchers’ throwing mechanics differ. Sixty youth baseball players with a mean age of 15 years old participated in a motion capture assessment with force plates installed in the mounds.
Each athlete threw three different pitch types classified as either fastballs, breaking balls, or changeups. The designation of pitchers vs. non-pitchers was determined by the players, parents, and coaches based on athletes who pitched often or who pitched very rarely.
The authors compared eight kinematic and three kinetic measures between 20 non-pitchers and 40 pitchers. They found that based on their chosen level of statistical significance, only maximum ground reaction force and maximum torso rotation velocity were significantly different between pitchers and non-pitchers, with pitchers exhibiting more of both metrics than non-pitchers. Both effect sizes were moderate.
Despite the difference in experience between the groups, there was barely any different in mechanics. This is both interesting and more or less supports the idea that youth athlete’s don’t need to be pitching before they are in high school to have the capacity to become pitchers later on in their career. These findings won’t necessarily change the way that we train youth baseball players but if interesting nonetheless. It would be interesting to test a wider range of ages and development levels in baseball to understand 1. if the mechanics of pitchers and non-pitchers begin to diverge (become more different) at higher levels of play and 2. what level of play does that start to happen. We might just have to add it to our study docket!
Can Machine Learning with IMUs Be Used to Detect Different Throws and Estimate Ball Velocity in Team Handball?
Workload monitoring in baseball is becoming more popular as groups create new ways of capturing proxies for training and in-game stimulus including the use of ball tracking technologies that measure proxies for intensity like ball velocity. The other primary way for measuring workload that we have discussed in the past is the use of sensors such as the MotusTHROW.
The ball tracking technology solution is less viable in handball. Throws don’t always happen in the same location, making sensor-based workload monitoring increasingly important in handball. Researchers at Nord University in Norway experimented with sensor data and machine learning to try and predict velocity for workload monitoring.
The idea is that if they can predict velocity with the sensor data, then they have a good enough measure of throw intensity—which can be used to estimate a cumulative workload.
The group tested seventeen handball players executing two different throw types with several different throw approach strategies, including standing, running, or jumping approaches. They placed an IMU on the participants’ arms, measuring the ball velocity of the throws using a radar device. Four different machine learning models were created using the IMU data to predict throw type, approach type, and ball velocity. The models were able to identify the throw type and approach type with between 80 and 90% accuracy and the ball velocity with an average absolute error of between 1 and 1.5 m/s, with a mean velocity around 21 m/s.
The value here for handball players and coaches is not necessarily to accurately predict velocity or count how many different types of throws each player executed during training or in the games (though that could be valuable). Instead, it’s to have a good enough proxy to estimate the intent of each throw. Estimation of intent can be used in conjunction with throw count to come up with a workload value that can be monitored throughout training and in-game performances.
Although this study investigated a different sensor system than MotusTHROW and they weren’t looking specifically at elbow torque, this study highlights the feasibility of using sensors to paint a workload picture for coaches and athletes that can theoretically be used for improving performance and managing player health.
For more information on how we use sensors to manage athlete workload, see our blog on it.