Public defence, Computer Science, MSc Nguyen Luong

Title of the thesis: Behavioral Sensing for Routine Characterization and Mental Health

Thesis defender: Nguyen Luong 
Opponent: Professor Albert Ali Salah, Utrecht University, Netherlands
Custos: Professor Petri Vuorimaa, Aalto University School of Science

Smartphones and wearable devices continuously produce information about how people move, sleep, communicate and use digital services. Behavioral sensing refers to the passive collection and interpretation of these digital traces to understand patterns of everyday behavior. This doctoral thesis examines how behavioral sensing can be used to study daily routines and their connections with mental health and well-being in everyday life. 

The study investigated two key characteristics of daily routines: stability and adaptation. The results show that people have individually distinctive “routine signatures” that remain relatively stable over time and across several areas of everyday behavior. These signatures reflect recurring combinations of behaviors such as activity, sleep, communication and device use. At the same time, routines are not fixed. Studies around the COVID-19 pandemic showed that movement and sleep patterns changed as restrictions and everyday circumstances changed, with systematic differences between demographic groups. 

The thesis also examined whether behavioral sensing can complement traditional questionnaires in mental health research. Studies involving people with depression and healthy participants found differences in mobility and communication patterns between the groups. Changes in behaviors such as time spent at home, phone use and communication were also associated with changes in depressive symptoms. A separate longitudinal study found that different patterns of internet use were associated with perceived stress. 

In addition to these empirical findings, the thesis makes practical contributions to behavioral sensing research. It presents an open-source software package for processing and analyzing multimodal behavioral data and evaluates how different technologies, including wearable and contact-free devices, measure sleep. These contributions aim to make digital behavioral studies more transparent, comparable and reliable. 

Overall, the findings show that behavioral sensing can provide a continuous view of everyday behavior that is difficult to capture with occasional questionnaires alone. The research provides new knowledge about both the persistence and adaptability of daily routines and shows how these patterns can be linked with psychological well-being in real-world settings. 

Behavioral sensing could support future research on mental health, personalized health monitoring and the detection of meaningful changes in everyday behavior. However, the observed relationships do not by themselves establish cause and effect. Behavioral sensing should therefore be seen as a complement to questionnaires and clinical assessment rather than as a replacement for them.

Key words: behavioral sensing, routines, mental health, computational social science

Thesis available for public display 7 days prior to the defence at Aalto University's public display page. 

Contact Information: 
Email: nguyen.luong@aalto.fi 
Linkedin: www.linkedin.com/in/khoinguyenluongnguyen