Development of Integrated Tools for Biomechanical Analysis in Sport Performance: Application to Cycling

Virtual models are adopted in every field of science for the description of real-world systems. Without a doubt, kinesiology is a challenging and fascinating playground where model developers can play a major role. The recent improvements in computational capabilities and data availability have triggered for a wide distribution of data that need virtual models to be processed and discussed critically. This thesis aims at developing visrtual models for the description of a real world system such as a cyclist. This system is broke down in three interdependent, intradependent and interlocked blocks: the bioenergetic (that links the mechanical and the metabolic power that the cyclist can supply), the biomechanical (that links the mechanical power output cranked by the cyclist and the contributions of the body actuators), the locomotion (that links the delivered mechanical power and the longitudinal speed of the centre of mass of the system). At the bioenergetic levels the focus was on the models for the prediction of the pulmonary oxygen consumption, the blood lactate concentration and the anaerobic sources depletion. The study revealed that the literature lacks of accurate models for the prediction of the slow component of the pulmonary oxygen consumption and for the accurate prediction of the blood lactate concentration during a generically shaped exercise. The literature has been revised and some models amended to best fit the experimental data: different solution have been included and the existing models have been improved. These models written with convenient mathematical form have been adapted as best suited for the solution to the high-intensity training design. An optimal control algorithm was asked to find the best solution for the different variables describing a training protocol. The application provided more time-efficient protocol to be validated in further observational studies. At the biomechanical level the focus was on the application of the optimal control algorithm to the solution of the predictive dynamics. A virtual model has been developed to fit in this framework and to best replicate the major features of the motion of a pedalling cyclist. Without any a priori experimental data it was possible to generate, to the best of our knowledge, the first predictive simulations of the biomechanics of sub-maximal cycling. These simulations found an application into the old debate about the pedalling technique that can provide the best efficiency and efficacy. The simulations suggest that the classic pedalling technique is adopted because the architecture and anatomical characteristics of the musculoskeletal system of the human body, rather than because adaptation to the repeated movement. The locomotion level of analysis has been the most explored by sport scientists. The available literature can be used to develop new and more accurate models of the locomotion of the cyclist-bicycle system. In this thesis the aim was to join bioenergetic model (the supply side) to the equation of motion of the centre of mass of the system. An optimal control algorithm has been used in three different situations: into the calibration of the parameters of the system that can best fit the experimental measures, into the solution of the pacing strategy problem, into the filetring and conditioning of outdoor cycling data of power, altitude and GPS positioning. The study revealed how optimal control and virtual models can be used for the solution of the pacing strategy with given environmental conditions and to best adapt outdoor data for further operations and applications. The diffusion of cheap measurement systems (e.g. GPS, powermeter), the spread of highly computational efficient CPUs on personal computers, the world wide access to the databases through the web, the neverending improvements in memory technologies and capacities (e.g. remote databases, clouds): in the future there will certainly be the need for a rapid, accurate and efficient handling of this big data. Many times it happens too fast, and the available data is bigger than what we can actually process and understand. Sport science is directly involved in this change. This study is not intended to be exhaustive, but wanted to provide new theoretical basis and mathematical integrated tools to help scientific community in filling the gap that technology advancements are creating between the available data and the information that we can all use.