January 29, 2017

iOS for Sports Performance Analysis and Training: a Laboratory in your Pocket

Co-written with Carlos Balsalobre-Fernandez


This post is about our collaboration to the development and validation of iOS apps for sport performance analysis. These four apps are cheap, easy to use, only require an iOS device (preferably with slow motion camera) and integrate published algorithms for biomechanical analysis of human movement. They have all be presented and compared to reference methods in published scientific studies.


Introduction

Being efficient on a sports field, on a paved road or a rugged path will depend on the quality of your sleep, your emotional state or what you eat or drink, but all things being equal, physiological and biomechanical factors will also influence your performance. Good news, we've some cheap and valid apps that will help you better identify and improve these performance factors. While heart rate and running or pedaling speed were often the only « physiological » variables easily accessible, the advent of GPS receivers and smartphones has opened a whole new dimension. Professional and amateur athletes will soon have the equivalent of a sport science laboratory in their pocket.... for those who are interested in further discovering and studying their capabilities. Fortunately, digital-free outdoor sport will still be possible!

The improvement of the muscular qualities involved in "explosive" disciplines such as jumping or sprinting is a crucial factor for training. Those who believe that the physical qualities do not explain the overall performance in a sport like football or rugby are right! But jumping a little higher or accelerating a little harder doesn't hurt, ask Cristiano Ronaldo if his phenomenal vertical leap did not help him to win a few important matches... Similarly, the French Rugby 7 team would probably not perform the same without his dragster Terry Bouhrawa and his blazing acceleration. A pure sprinter will never be a good football or rugby player without years of training, but between two good team sports players, the one who can jump higher or run faster will certainly have an advantage in the game. Citius = Altius = bonus!

In his masterpiece "Le mouvement" (published in 1894), Etienne-Jules Marey, considered as the first exercise physiologist and biomechanist, showed his analysis of human and animal movement. His "chronophotography" method allowed him to study many movements including jumping and sprinting, using photography tools with series of images, approaching the rate of 20 to 24 frames per second. Beyond advancing knowledge about human movement with this technical innovation, Marey is considered to be one of the precursors of the scientific film, or even the film itself.

Etienne-Jules Marey (Beaune, 1830 - Paris, 1904), doctor and French physiologist, successor of Claude Bernard at the National Academy of Sciences, of which he was President, was one of the precursors of the scientific movement analysis. He invented photographic and cinematographic tools and methods for the observation of moving bodies. Most of his work on this theme are exposed in his major work: « Le Mouvement ». In some excerpts from a letter written to his mother, he recounted the issue of the accuracy of the measurements, but also administrative and technological constraints to the development of knowledge on human movement. This text has not aged a bit!

Latest Apple (tablets and phones) devices allow video measurements at 240 fps. About 10 times faster than the revolutionary devices designed by Etienne-Jules Marey at the end of the XIXth century. This rate of 240 fps is similar or even higher than that of cameras used in the laboratory for the analysis of human movement and sports. Indeed, many publications in Biomechanics analyzed jumping, walking or running based on expensive cameras at frequencies of 100 to 200 fps. Happy coincidence, this advanced technology comes at the same time as the publication and validation of mathematical approaches to analyzing determinants of sport performance on the basis of simple parameters: jump height, time of contact with the ground or running speed during a sprint (see below). This technology-science context gave birth to cheap iPhone and iPad apps (less than 10 USD), that have been but scientifically validated! Increasingly popular in sports structures, they enable lower cost, measurements of variables of the sports performance far much more complicated to implement. One of their advantages is that they perform many calculations in a fraction of a second. A laboratory in the pocket of the coaches, and athletes of all levels...

1/ Jump performance and force-velocity-power profile with MyJump


The work carried out by our team allowed these past years to validate a complete method of analysis of the force-velocity profile and performance during short, maximal impulse ballistic movements (e.g. vertical jump or a first step during acceleration) on the basis of the simple measure of jump height (all papers here). Performing a few loaded vertical jumps allow to know the individual profile of the athlete, compare it to its optimal profile (i.e. the profile with which jump height will be maximized, for a given level of maximal power output), and in turn specially to design an effective and individualized training. A kind of personalized prescription. With MyJump, jump height is measured simply by filming the athlete's feet. By watching the jump frame by frame, tap on the screen when the feet leave the ground, and again when the feet touch the ground on landing a few tenths of a second later. The application basically counts the number of frames between the two events, computes the time spent in the air, and the laws of falling bodies do the rest, thank you Galileo and Newton. Simplicity is the ultimate sophistication.
As body mass is entered in the user profile, the app calculates the maximal power of the athlete, his/her force-velocity profile and optimal profile with a few clicks on the screen, just by making a few additional jumps (see method here) with different levels of load on the shoulders. As in the illustration, the individual force-velocity deficit informs on training lead (physical capability to develop, magnitude of the deficit, etc.) and repeating the test over the training period allows a very easy follow-up. The French volleyball federation and many clubs of volleyball or basketball are currently using the application in the monitoring of the physical qualities of the players, and the individualization of training program (see here).
This app is so simple and cheap that a rigorous proof of validity was necessary to reassure researchers and coaches who sometimes think that a good tool is necessarily costly, and vice versa, a "gadget" for iPhone can not be reliable. This validity has been tested by Carlos Balsalobre-Fernandez and his collaborators who have compared the jump height given by the app to that calculated by the Rolls Royce of biomechanics measurement devices: the force plate. The findings show a very high concurrent validity and reliability of measures (paper here), which was also mentioned in the very serious British Journal of Sports Medicine (here). In addition, several other studies have replicated the measurements and clearly confirmed this validity…

Figure 1. Left : with MyJump, you just need to film the feet of the athlete in slow motion and tap on the screen at take off and landing, easy. Middle : after a few additional loaded jumps, the app displays the force-velocity profile of the athlete, in comparison to his optimal profile, i.e. the profile with which jump height will be maximal, for the level of maximal power of the athlete (51.8 /kg here). This profile is computed according to Samozino et al.'s method (see here). Right : then, the app displays a training program that should be followed in order to reduce the force-velocity imbalance (gap between actual and optimal profiles) and in turn improve jump height (see this study for details). Next step: training !    

2/ Sprint acceleration performance and force-velocity-power profile with MySprint

Now that you have improved your jump performance, let’s check your sprint acceleration. Using the same simple idea (i.e. filming sports movement in order to identify the key elements by watching the slow motion video), MySprint app will ask you to film a 30-m standing start acceleration, and then touch the screen when the athlete’s body crosses markers set every 5 meters. Then, the app calculates sprint performance and split times, but also computes their production of force onto the ground and external power, as well as the effectiveness of their ground force application (i.e. how horizontally their push is oriented…the more horizontal the better). These biomechanical variables explain a major part of acceleration and overall sprint performance (see this study), and their easy and accessible measurement is a valuable aid for training, on an individual basis. For instance, for a same given performance over 20 meters, we observed for that some international Rugby 7s players had an effective ground force application but lacked overall strength, while others had impressive muscle power but a very inefficient push on the ground. This information is helpful to guide the training of each player according to his/her own needs, and profile. The calculations involved here, which are included in the app, also use the laws of Newtonian mechanics applied to the body of the athlete (details here), and only require to know the body mass of the subjects, and either their 5-m split times over a 30-m, or their running speed as a function of time (e.g. with a radar).


Figure 2. Left: with MySprint, the athlete is filmed in slow motion, so that 5-m split times are calculated when you tap on the screen as the athlete’s body crosses each 5-m pole. Right: then, from the 5-m splits, the app computes the main mechanical outputs during the acceleration using the method validated by Samozino et al. (see here): horizontal net force, velocity, power, and the effectiveness of ground force application (i.e. the horizontal orientation of the ground reaction force. As for MyJump, data are stored in the athlete’s profile, and may be sent by email for deeper analysis and individualized training.



These approaches using individualized strength-speed-power training have been recently discussed (here), and as for MyJump, MySprint validity has been tested by comparison with reference devices for sprint assessment: photocells and radar. As for MyJump, the results show a very good agreement (article here).

3/ Analyzing and monitoring running biomechanics with Runmatic

For those who prefer running from longer durations, an application developed on the same principle (Runmatic), allows a biomechanical analysis of the running pattern. The latter is often described as a "spring-mass" system: the rider is mechanically described as a mass bouncing on a spring. As simple as it may seem, this model correctly describes the overall mechanical behavior of the human body during running. Based on the measurement of ground contact and flight time (obtained from slow-motion video recording…), it is possible to calculate the main features of the spring-mass system: peak force applied onto the ground, vertical displacement of the center of mass during the support phase, and musculo-tendinous stiffness of the lower limb (see method here). These variables, as well as the differences and asymmetries between right and left leg were usually measured on costly instrumented treadmills or force plates, or with relatively heavy to implement and/or costly field tools such as opto-electronic rails. From the video recording of the runner (see figure), Runmatic allows coaches and researchers to calculate all these variables and follow their changes over time, with training, or even with the development of running-related injuries. Same procedure: a slow motion video, a few taps on the screen when the feet come in contact with the ground and take off, et voilà! As for the other apps, a validation has been performed in comparison to reference devices for contact and flight time measurement (article here).


Figure 4. Left : you need to film the runner's feet and identify contact and take-off instants for eight consecutive steps. Middle : Runmatic then computes the main mechanical variables of the running pattern, for each leg, and the asymmetry between legs (contact time in the example here). Right : a summary of data is displayed, and changes over time may be followed as the app calculates the relative changes between two consecutive measurements.


One of the most interesting modules of the app allows a follow-up and the immediate display of changes in mechanical variables, to easily follow the effect of training or rehabilitation. This can also help the runner of coach/physical therapist to detect an unexplained change in the symmetry between the two legs (often seen in the context of lower limb pathologies like  tendinitis or periostitis). And this even before the runner is conscious of any pain associated with a potential pathology. As natural and easy as this activity may seem, running mechanics is precision engineering...

4/ Measuring your maximal strength (1RM) with PowerLift

 Decidedly very active, our Spanish colleagues finally focused on another classic of the training toolbox: the calculation of the maximum load an athlete can lift during resistance exercises (or 1RM). This value represents the load that can be lifted only once, and knowing it allows athletes to estimate their « personal best », but also to design their training programs accordingly. The reference method to determine this load is pretty obvious: try increasing resistance loads until you can no longer push... but in practice this means many efforts, time and fatigue. Other methods have been proposed to estimate the 1RM by knowing the maximum number of repetitions for a given load. For example, if I can lift 10 times (and not 11...) a bar of 60 kg at the bench press, then I can estimate my 1RM. This indirect approach is convenient for a quick estimate, but they give a very rough value, since the regression equations on which they are based does were obtained in specific populations.
The last of these applications uses the principle of the video capture of the displacement of a bar or body during traditional strength and conditioning exercises. Based on the number of images (hence time) flowing between the beginning and the end of the thrust, Powerlift calculates the average speed of ascent of the bar/body. Then, as MyJump, by performing several tests with sub-maximal loads, the application calculates the maximum load that may be lifted (at an extremely low speed). This approach is based on the fact that the speed at which each of us "pushes" his own 1RM is fairly stable and common to all the athletes, while the load itself is very variable. The app determines the 1RM value according to the speed of the movement, and not the load (see for instance here).
The data computed by the app have been recently shown reliable in comparison to reference devices (article here).


Figure 5. Left : lift four sub-maximal loads, as fast as possible, and then from the slow motion video recorded, Powerlift will estimate your 1RM. Right : the computation (99.6 kg here) is based on the loads lifted, and the identification of the distance and time of motion. That's it.


5/ Applications and limitations



The main advantage of the tools described here is that they present a very effective combination between cost (a few euros to add of course to the iPhone/iPAd needed), ease of use, quality of measures and scientific basis of the theoretical concepts used. In practice, what coaches and physical trainers are looking for is accuracy and relevance of the parameters analyzed, but also and more importantly their accessibility. Key physiological and biomechanical variables have long been only accessible to « happy fews » who had access to the expensive equipment and associated expertise. The apps presented in this article bring a change of perspective, accurate doesn't necessarily mean complex and costly, and relevant doesn't mean inaccessible.

The main limitation of these applications is that they are (for now) available only on Apple iOS systems, and this for two main reasons. First of all, all the latest iPhone and iPad devices feature slow motion video recording at 240 fps, which determines the precision of the measurements and calculations. This is not the case of other smarthpones/tablets currently on the market. Then, being developed by sport scientists and « amateur » coders in an independent manner, the Apple environment ensures the stability of these applications on all Apple devices, which would be much more complicated on other platforms (hundreds of different devices running Android for example). By the way, this strongly decreases the risk of piracy. Nevertheless, the scientific basis of the measures being published and available, anti-Apple geeks are free to develop equivalent apps, provided video sensors allow equivalent precision.

Finally, as for all new performance tools that flourish these days, their main advantage turns very often into a major limit: ease of access to the data. It becomes so simple to measure biological variables that you eventually accumulate them and store them without even knowing what they "say", and understanding how the information they carry may be used in practice.
Quantifying objectively, precisely and simply athletes’ physical qualities is paramount, but this does not simplify the complexity of neurophysiological and biomechanical adaptations hidden behind all these values.

In summary, the three benefits of these innovative apps in the field of sports performance assessment are their ease of use and accuracy, compared to reference methods, their validity supported by published studies, and the fact that they include the published processing methods. They are already used on a regular basis in many sports structures, rehabilitation centers or even research laboratories. Simple, practical and reliable doesn't necessarily mean expensive...

"Simplicity is the ultimate sophistication"   
Leonardo da Vinci