MSc Thesis Defense by Adam Honore
Understanding Factors Influencing Loneliness And Corresponding Quality Of Life via Interviews, Smartphone Sensing And Data Modeling
Loneliness is a widespread phenomenon that affects people all over the world, in all layers of society. It has been shown to increase the likelihood of premature mortality by as much as 26%, which conveys the importance of being able to assess and predict loneliness. In this thesis, we gather data on smartphone usage patterns and assessed loneliness from three students over a period of four weeks. Smartphone usage features are extracted from the data of one subject and analyzed using data modeling and machine learning, to delineate the relationships between these features and the assessed loneliness. Predictive models have been evaluated, and the best performing model has been selected. Our predictive model shows signs of overfitting but is still able to confirm a link between an increase in physical activity and decreased loneliness, as is also found in related studies. Additionally, we also found links between increased amounts of time spent near other people, as measured by the subject’s semantic location, and decreased loneliness. A link between an increase in the number of times that the subject used their Email app and decreased loneliness was also found. Based on our work in this thesis we experimentally validate theories from related studies and lay the foundation for future work in this area.
Supervisor: Katarzyna Wac (remotely from Stanford office)
Censor: Tomas Sokoler, ITU