With his French Open win earlier this month―his 19th career Grand Slam―Novak Djokovic now trails Roger Federer and Rafael Nadal by only one for the all-time Slam record. And now, going into Wimbledon, which begins today, the world No. 1 and the overall favorite is in prime position to pull even with his rivals, especially given that Nadal has pulled out to rest his body and Federer has not won an ATP title since 2019. Still, grass is Federer’s best surface, and Wimbledon is his best Slam. He’s won eight titles at the All-England Club.
Meanwhile, home-crowd favorite Andy Murray is returning to his first Wimbledon since 2017. And then there is the rising class of young stars―Daniil Medvedev (French Open quarterfinalist), 25; Stefanos Tsistipas (French Open runner-up), 22; Alexander Zverev, 24 (French Open semifinalist); and Matteo Berretini, 25 (French Open quarterfinalist)―who seem ready to break through the Big 3 monopoly at any moment. Simulating the tournament using adjusted Elo ratings, I calculated the estimated probabilities for every player in the draw to not only reach any given round but also to win it all.
… the thinking is changing. Even the sports medicine doctor who encouraged icing for athletic injuries decades ago as part of his RICE method (rest, ice, compression and elevation) has backed away from cold treatment.
So why is that, you ask? Basically, icing your sore muscles puts the freeze on your body’s natural – and highly effective – healing response.
Large high-quality datasets of human body shape and kinematics lay the foundation for modelling and simulation approaches in computer vision, computer graphics, and biomechanics. Creating datasets that combine naturalistic recordings with high-accuracy data about ground truth body shape and pose is challenging because different motion recording systems are either optimized for one or the other. We address this issue in our dataset by using different hardware systems to record partially overlapping information and synchronized data that lend themselves to transfer learning. This multimodal dataset contains 9 hours of optical motion capture data, 17 hours of video data from 4 different points of view recorded by stationary and hand-held cameras, and 6.6 hours of inertial measurement units data recorded from 60 female and 30 male actors performing a collection of 21 everyday actions and sports movements. The processed motion capture data is also available as realistic 3D human meshes. We anticipate use of this dataset for research on human pose estimation, action recognition, motion modelling, gait analysis, and body shape reconstruction. [full text]
Pascal Fua, Professor, Computer Vision Laboratory, EPFL
Title: Athletic 3D Pose Estimation
Abstract: While supervised body pose estimation is rapidly becoming a mature field, the bottleneck remains the availability of sufficiently large training datasets, which cannot be guaranteed for many kinds of human motions. An effective way to address this is to leverage unsupervised data to learn a low-dimensional representation of poses. Then, it only takes very little annotated data to train a regressor to predict 3D poses from this representation. In this talk I will discuss several classes of techniques that rely on unsupervised learning objectives that exploit the availability of multi-view footage and of video sequences for this purpose. I will demonstrate the benefits of this approach to track skiers, divers, and ball players.
… we are dazzled by fancy graphs and neatly organized charts. We’re fooled into thinking that the mere possession of data will help us better train our athletes. We hope that our “experience” will help us know what to do, or that the nebulous concept of “actionable information” will reveal itself. Our sophisticated data collection lulls us into thinking that we are being scientific, but really we are not.
… One of the key components of most deep learning–based computer vision applications is the convolutional neural network (CNN). Invented in the 1980s by deep learning pioneer Yann LeCun, CNNs are a type of neural network that is efficient at capturing patterns in multidimensional spaces. This makes CNNs especially good for images, though they are used to process other types of data too. (To focus on visual data, we’ll consider our convolutional neural networks to be two-dimensional in this article.)
Every convolutional neural network is composed of one or several convolutional layers, a software component that extracts meaningful values from the input image. And every convolution layer is composed of several filters, square matrices that slide across the image and register the weighted sum of pixel values at different locations. Each filter has different values and extracts different features from the input image. The output of a convolution layer is a set of “feature maps.”
When stacked on top of each other, convolutional layers can detect a hierarchy of visual patterns. For instance, the lower layers will produce feature maps for vertical and horizontal edges, corners, and other simple patterns. The next layers can detect more complex patterns such as grids and circles. As you move deeper into the network, the layers will detect complicated objects such as cars, houses, trees, and people.
A team of Virginia Tech researchers from the Department of Mechanical Engineering and the Macromolecules Innovation Institute has created a new type of soft electronics, paving the way for devices that are self-healing, reconfigurable, and recyclable. These skin-like circuits are soft and stretchy, sustain numerous damage events under load without losing electrical conductivity, and can be recycled to generate new circuits at the end of a product’s life.
Numerous researchers have leveraged publicly available Internet sources to publish publicly obtained data (POD) studies concerning various orthopaedic injuries in National Football League (NFL) players. Purpose:
To provide a comprehensive systematic review of all POD studies regarding musculoskeletal injuries in NFL athletes and to use anterior cruciate ligament (ACL) injuries in NFL players to quantify the percentage of injuries identified by these studies. Study Design:
Systematic review; Level of evidence, 4. Methods:
A systematic review was conducted to identify all published studies utilizing POD regarding ACL injury in NFL athletes from 2000 to 2019. Data regarding player demographics were extracted from each publication. These results were compared with prospectively collected data reported by the teams’ medical staff to the NFL Injury Surveillance System database linked to the League’s electronic health record. An ACL “capture rate” for each article was calculated by dividing the number of ACL injuries in the POD study by the total number of ACL injuries in the NFL injury database occurring in the study period of interest. Results:
A total of 42 studies were extracted that met the definition of a POD study: 28 evaluated a variety of injuries and 14 dealt specifically with ACL injuries, with 35 (83%) of the 42 studies published during or since 2015. POD studies captured a mean of 66% (range, 31%-90%) of ACL injuries reported by the teams’ medical staff. This inability to capture all injury rates varied by position, with 86% capture of ACL injuries in skill athletes, 72% in midskill athletes, and 61% in linemen. POD studies captured 35% of injuries occurring during special teams play. Conclusion:
The frequency of studies leveraging publicly obtained injury data in NFL players has rapidly increased since 2000. There is significant heterogeneity in the degree to which POD studies correctly identify ACL injuries from public reports. Sports medicine research relying solely on publicly obtained sources should be interpreted with an understanding of their inherent limitations and biases. These studies underreport the true incidence of injuries, with a bias toward capturing injuries in more popular players.
Context: Return to activity(RTA) assessments are commonly administered following ACL-Reconstruction(ACLR) to manage post-operative progressions back to activity. To date, there is little knowledge on the clinical utility of these assessments to predict patient outcomes such as secondary ACL injury once returned to activity.
Objective: To identify what measures of patient function at 6-months post-ACLR best predict return to activity and second ACL injury at a minimum of 2-years following ACLR.
Design: Prospective-cohort
Setting: Laboratory
Patients: A total of 234 patients with primary, unilateral ACLR completed functional assessments at approximately 6-months post-ACLR. A total of 192(82%) completed follow-up ≥ 2-years post ACLR.
Main Outcome Measures: Six-month functional assessments consisted of patient reported outcomes, isokinetic knee flexor and extensor strength, and single-leg hopping. The ability to return to activity and secondary ACL injury were collected at a minimum of two-years following ACLR.
Results: In patients who did RTA(n=155), a total of 44(28%) individuals had a subsequent ACL injury; graft n=24(15.5%), contralateral ACL n=20(13%). A greater proportion of females had a secondary injury to the contralateral ACL(15/24, 63%) whereas a greater proportion of males reinjured the ipsilateral ACL graft(15/20, 75%, P=.017) Greater knee extension symmetry at 6-months increased the probability of reinjury(B=.016, P=.048). In patients who RTA before 8-months, every 1% increase in quadriceps strength symmetry at 6-months increased the risk of reinjury by 2.1%(B=.021, P=.05). In patients who RTA after 8-months, every month that RTA was delayed reduced the risk of reinjury by 28.4%(B=−284, P=.042).
Conclusions: Patients with more symmetric quadriceps strength at 6-months post ACLR were more likely to experience another ACL rupture, especially in those who returned to sport earlier than 8-months after the index surgery. Clinicians should be cognizant that returning high functioning patients to activity earlier than 8-months post-ACLR may place them at an increased risk for reinjury. [full text]
Three decades after its founding, the Knight Commission on Intercollegiate Athletics now sees the college sports world embarking on the “transformational change” it has long advocated for. The change just happens to be driven by an issue the commission has been reluctant to support: college athletes profiting while playing.
This paradox finds the Knight Commission scrambling to establish credibility on athlete rights while being criticized by player advocates as, at best, ineffectual, and at worst, a fig leaf for the financially exploitative status quo.
“I don’t even consider the Knight Commission a reform organization,” said Ramogi Huma, executive director of the National College Players Association, which was instrumental in crafting California’s pioneering name, image and likeness bill. “To me, a reform organization goes and fights. I have always considered the Knight Commission a think tank… a shadow of the NCAA.”
Elite football players need to handle enormous physical and physiological demands during matches. Therefore, in the scientific literature there are many studies which evaluate how different load variables respond to football players during competition, such as lactate levels, heart rate or the effort exerted.
In recent years, due to technological advances and especially to the development of the global positioning systems (GPS), the coaching staff can painstakingly analyse the physical (or mechanical) demands of the players. These and other systems such as accelerometers give information on variables such as the distance covered or the amount of hard effort made above a certain speed, which can be indicative of the “hardness” of the game.
Several studies have tried to analyse through the GPS the demands that are presented in football players during a match. On the one hand, these studies can serve to discover if these demands are dependent on the type of game or player. Moreover, this information is commonly used to design training sessions that simulate the demands that players will have to cope with in the competition. In this sense, most of the studies are focused on analysing the “worst-case scenario”, i.e., the biggest physical demands that players have to cope with during a game.
