Visual Analytics for Understanding and Predicting Depression Mechanisms

Master Thesis


Collaborating experts from the domain of behavioral sciences and neuropsychology are interested in investigating whether depressed people differ from healthy people, based on available brain data, physiology, self-reports, and their behavior [1]. To this end, they are conducting a study with measures that are mostly focused on tracking the process of rumination, which is a key factor of depression. To further understand rumination and depression, our collaborating domain experts are observing the status of depressed patients and try to identify and predict whether specific interventions can perturb it, as well as the effect of this perturbation [2].

Visual analytics for cohort exploration and prediction (see [3-5]) have not been employed yet for depression analysis. In this project, we would like to design and develop in close collaboration with our collaborating domain experts a visual analytics system that facilitates and supports the exploration, analysis and comparison of data streams belonging to depressed vs. healthy people. The challenge in this project is that the available data are multi-variate and multi-timescale, i.e., occur on many different time scales. In addition to that, domain experts do not even know or anticipate relationship between many of these variables.


We will investigate appropriate solutions for the visual exploration, analysis, and comparison of relationships between multi-variate and multi-timescale data streams belonging to healthy vs. depressed participants of a previously conducted clinical study. In particular, we will focus on the following tasks:

1) Understand the inter-relationships between variables at different timescales (e.g., EEG measurements (at millisecond resolution), heart rate (and its variability, at second resolution), behavior (at multi-second resolution) and self-report questions (every few hours)). This will help us understand how these variables can index depression and rumination, and how they can complement each other at various timescales.

2) Understand and track the effect of repeating (at different timescales) measurements (e.g., a self-report occurs 10 times a day, while a short cognitive task happens only once in a day), as well as the time-dependent relationship between variables?

3) Understand and identify how the available multi-variate, multi-timescale data can be employed to predict a mental and physical state in the future (e.g., what patterns of past data can tell us how the participant’s mental and physical state will be in the future?)


  • Interest and previous knowledge in visual analytics and/or visualization.

  • Interest in visualizing multi-variate, multi-timescale cohort data.

  • Basic machine learning knowledge (for the prediction part, see Task 3).

  • Good programming skills.

  • Creativity and enthusiasm.

  • Willingness to collaborate within a multi-disciplinary setting.


To be discussed (depending on the background of the student).


[1] Jin, C.Y., Borst, J.P. and van Vugt, M.K., 2019. Predicting task-general mind-wandering with EEG. Cognitive, Affective, & Behavioral Neuroscience, 19(4), pp.1059-1073.

[2] Targum, S.D., Sauder, C., Evans, M., Saber, J.N. and Harvey, P.D., 2021. Ecological momentary assessment as a measurement tool in depression trials. Journal of psychiatric research, 136, pp.256-264.

[3] Furmanova, K., Muren, L.P., Casares-Magaz, O., Moiseenko, V., Einck, J.P., Pilskog, S., and Raidou, R.G., 2021. PREVIS: Predictive visual analytics of anatomical variability for radiotherapy decision support. Computers and Graphics (Special Section on Visual Computing in Biology and Medicine), vol. 97: 126-138.

[4] Bernold, G., Matkovic, K., Gröller, M.E., and Raidou, R.G., 2019. preha: Establishing Precision Rehabilitation with Visual Analytics. Eurographics Workshop on Visual Computing for Biology and Medicine (EG VCBM), pp.79-89.

[5] Blumenschein, M., Behrisch, M., Schmid, S., Butscher, S., Wahl, D. R., Villinger, K., Renner, B.R., Reiterer, H., & Keim, D. A. (2018). SMARTexplore: Simplifying High-Dimensional Data Analysis through a Table-Based Visual Analytics Approach. IEEE Conference on Visual Analytics Science and Technology (VAST), vol. 1: 36-47.


This project is done in collaboration with the University of Groningen and the University Medical Center Groningen, the Netherlands, together with Marieke van Vugt (Cognitive Modeling), Marlijn Besten (Clinical Neuropsychology) and Marie-José van Tol (Psychiatry).



For more information please contact Renata Raidou.