| Dozent/in |
Dr. Cristina Ehrmann |
| Veranstaltungsart |
Vorlesung |
| Code |
HS261691 |
| Semester |
Herbstsemester 2026 |
| Durchführender Fachbereich |
Gesundheitswissenschaften |
| Studienstufe |
Master |
| Termin/e |
Do, 17.09.2026, 14:15 - 16:00 Uhr, HS 4
Do, 24.09.2026, 14:15 - 16:00 Uhr, 3.B47
Do, 01.10.2026, 14:15 - 16:00 Uhr, HS 4
Do, 08.10.2026, 14:15 - 16:00 Uhr, HS 4
Do, 15.10.2026, 14:15 - 16:00 Uhr, HS 4
Do, 22.10.2026, 14:15 - 16:00 Uhr, 3.B47
Do, 29.10.2026, 14:15 - 16:00 Uhr, HS 4
Do, 05.11.2026, 14:15 - 16:00 Uhr, HS 4
Do, 12.11.2026, 14:15 - 16:00 Uhr, HS 4
Do, 19.11.2026, 14:15 - 16:00 Uhr, HS 4
Do, 26.11.2026, 14:15 - 16:00 Uhr, HS 4
Do, 03.12.2026, 14:15 - 16:00 Uhr, HS 4
Do, 10.12.2026, 14:15 - 16:00 Uhr, HS 4
Do, 17.12.2026, 14:15 - 16:00 Uhr, HS 4 |
| Umfang |
2 Semesterwochenstunden |
| Inhalt |
Inferring associations from high-dimensional observational data is notoriously difficult. Graph neural networks (GNNs) are a type of artificial neural network that is designed to process data represented as graphs. In this course, students will learn to:
- use various machine learning methods (e.g. Lasso regression) to generate GNNs.
- estimate and visualise the GNNs for real-world data.
- understand the difference between the concepts of association and causation within GNNs.
- understand key centrality measures (e.g. degree centrality, between centrality) and explore their potential applications in the GNNs.
- identify the variable that "matters most" in the estimated GNN using simulation techniques.
- visualize and analyze the centrality measures in longitudinal GNNs.
The course combines lectures with practical exercises, enabling students to apply GNNs to real-world healthcare examples. |
| Schlagworte |
Gender/Diversity |
| Lernziele |
Students will be able to:
- build and visualize GNNs;
- visualize centrality measures for each variable and assess the stability of these measures;
- identify the variable that "matters most" in the estimated GNN using simulation techniques;
- present GNN results to a non-specialist audience. |
| Voraussetzungen |
This course requires basic knowledge of the R programming language. |
| Sprache |
Englisch |
| Begrenzung |
The course is limited to 12 participants. The limit will be administered via Moodle according to the chronological order of registration. From 31st August 2026, 12:00 p.m. (noon), it will be possible to register via Moodle. As soon as 12 participants are registered, the registration window will close automatically. |
| Anmeldung |
Moodle: https://elearning.hsm-unilu.ch/course/view.php?id=1086 |
| Leistungsnachweis |
Student performance will be evaluated based on the following:
- Regular completion of weekly hands-on exercises (30%); and
- The final project, which will be assessed on the quality of the work and the clarity of its presentation (70%).
Attendance is mandatory for all students.
IMPORTANT: In order to earn credits and participate at the exam registration via Uni Portal within the exam registration period is MANDATORY. Further information: www.unilu.ch/en/study/courses-exams-regulations/health-sciences-and-medicine/exams/ |
| Abschlussform / Credits |
Exercises, final project, attendance mandatory / 3 Credits
|
| Hörer-/innen |
Nein |
| Material |
The lecture slides, exercises/data, and the student presentations will be made available to all students. |
| Literatur |
Epskamp, S., & Fried, E. I. (2018). A Tutorial on Regularized Partial Correlation Networks.
Psychological Methods, 23(4), 617-634.
Robinaugh DJ, Millner AJ, McNally RJ. Identifying highly influential nodes in the complicated grief network. J Abnorm Psychol. 2016;125(6):747–57.
More literature references (mostly journal articles) will be provided during the lecture. |