Uncovering the relationship between mood and sport performance using context-aware mobile sensing

Abstract

Understanding how the changes in our mood have an impact on the events and experiences that shape our daily lives is an essential task in affective science. This fact becomes even more critical in the area of sports since mood is widely recognized as a contributing factor to determining the performance of athletes. As mood fluctuations are strongly influenced by the context surrounding us, identifying the situations and behaviors that trigger these mood changes is crucial to optimizing the performance of athletes during practices and competitions. Traditional research in this field has focused on studying the mood of athletes at specific points in time, primarily within controlled settings, and exploring its relationship with a specific sports outcome. However, little research has been oriented towards exploring the mood variations during the athletes’ daily lives, which is even more critical due to the dynamic nature of mood. This lack of longitudinal studies is partly caused by the limitations of existing data collection technologies and data analysis methodologies. In recent years, the emergence of mobile devices has enormously facilitated the task of continuously monitoring mood and context in free-living environments. However, there are still numerous challenges to be overcome when representing and analyzing an individual’s daily life and extracting valuable knowledge from that information. In this thesis, the use of mobile sensing and data science to explore the in-context mood fluctuations and their relationship with the performance of elite athletes is investigated. To that end, novel data collection systems and longitudinal analysis methodologies are proposed to overcome the limitations of traditional research approaches. In particular, a mobile-based monitoring platform is developed to continuously capture people’s mood and context during their daily lives. The platform, which is tested and validated, is employed to gather affective, behavioral, and contextual data in two longitudinal experiments. The primary considerations and issues related to the development of these experiments during extended periods in free-living environments are gathered and discussed, contributing to the future design of more effective longitudinal data collection studies. Moreover, an innovative methodology for analyzing longitudinal behavior-related data is proposed to provide a new approach for analyzing the between-person differences in affective behavior and the within-person fluctuations of mood based on the surrounding context. This approach proves to be effective in analyzing how the mood of athletes behaves during their daily lives and leads to preliminary conclusions about the effect of these mood fluctuations on the performance of the studied population of athletes. Through these contributions, this thesis aims to provide a complete picture of the process followed from the data collection to the extraction of interpretable information and offers a new perspective on how mood, behavior, and context can be analyzed together. Furthermore, its potential uncovers future research directions such as integrating new personal sensing devices for the unobtrusive collection of behavioral and contextual data, or extending longitudinal data analysis approaches to account for the temporal dimension of mood fluctuations.