Making big decisions about performance, injury prevention, selection and training in sport is exceedingly complex. While sports managers often possess large amounts of data to inform their choices, it is typically very large, noisy and sometimes patchy. Hence, they are left to rely upon intuition for decision making. But what if that data could be organised - through intelligent software that incorporates new machine learning and probabilistic reasoning approaches - to capitalise on this data and provide crucial strategic decision making advice for performance improvements? We have developed intelligent systems and software that meets these goals. The software solution primarily concerns decision making in regards to selection, training and competition strategies of elite athletes. A key factor is that the software provides information that is useable by the coach, sometimes even during the course of an athletic event. The software is currently primarily used to provide competition strategy advice for athletes involved in competitions involving multiple skills such as the cycling omnium, triathlon, swimming medley, decathlon, modern pentathlon, diving and gymnastics. Results to date show critical advice has been gleaned from the software. For example, in the cycling omium, we saw that whilst coaches and sport scientists see a major distinction between sprint and endurance events, the difference between time trials and mass start events - including Scratch race, Elimination Race and Points Race - is even greater. This specific information has an impact on selection and strategy in that sport. The software is also being used by the Victorian Institute of Sport to regulate training load. For each specific sport being modelled, the software uses data and optimization techniques to probabilistically find the best way for an athlete to achieve the required result, whether it be winning a medal or placing well in preliminary or intermediate contests. The software also provides users with a likelihood measure which indicates the probability of an athlete finishing the competition in a desired position. The information is provided in real time - for example useful information was provided to Australian Cyclists competing in the omnium at both the 2012 World Championships and the London Summer Olympic Games. After the elimination race (the third event in the omnium) cyclists were told what times they needed to achieve in the 1KM time trial and the 4 KM individual pursuit and how close they needed to finish in relationship to their closest competitors in the scratch race. Tri-athletes are told how close to their major competitors they need to be following the swimming and cycling legs of their event. Rowing crews are advised about what position they should be in at the conclusion of each 500 meters of a standard 2-kilometer event.
© Copyright 2014 The 3rd Wingate Congress of Exercise and Sport Sciences. Veröffentlicht von The Wingate Institute for Physical Education and Sport. Alle Rechte vorbehalten.