Quantitative psychology · data science · digital mental health
I am a quantitative psychologist and data scientist who develops methods to study complex psychological processes and their links to mental health. My work combines intensive longitudinal data, passive smartphone sensing, multimodal measurement, dynamic systems, and machine learning.
About
My methods research focuses on dynamical-system methods, control theory, network analysis, and machine learning. I use intensive longitudinal data—such as experience sampling, smartphone interactions, wearable signals, and Screenomics—to understand how people adapt to their environments and how psychological processes unfold over time.
My recent work integrates passive sensing data from smartphones and wearables, with an emphasis on developing interpretable behavioral measures for mental health and digital well-being.
Research
Digital phenotyping
I study how passively collected smartphone behavior can support measurement of depressive symptoms and digital well-being. In work with Mindstrong colleagues, we created interpretable behavioral features and used multilevel models to examine within- and between-person associations with depressive symptoms over time.
Network modeling
Emotions and social behaviors form interconnected systems. My work uses dynamical-system methods to model those networks and examine how mutually reinforcing temporal relations, such as positive feedback loops, relate to depression and anxiety.
Control design
I apply Boolean-network methods to model nonlinear psychological dynamics and to design control strategies that can help move a system away from undesirable states. The long-term aim is interpretable, person-specific intervention design across development and aging.
Screenomics & AI
Screenomics captures sequences of smartphone screenshots to study everyday experience. My work has examined how temporal features of visual stimulation—including color, text, and sentiment—help predict task switching on smartphones.
Selected publications
Resources
POMPOM on CRAN ↗ (Person-Oriented Modeling and Perturbation on the Model) combines structural vector autoregression and impulse-response analysis. It supports in-silico experiments and network-level metrics that map onto psychological concepts such as emotion-regulation efficiency.
The original tutorial pages are retained as links while their source materials are gathered for a future update.
Education
Contact
For research collaboration, data-science projects, or questions about these methods, reach me at vwendy@gmail.com.