The Computational Psychiatry Course (CPC, link to website) takes place each year in Zurich since 2012, and is opened to students and researchers from other universities since 2014. It is organised by the Translational Neuromodeling Unit, led by Klaas Ennos Stephan at the University of Zurich and ETH Zurich. The course takes place both online through Zoom and onsite at Zurich. I took this course as a postdoctoral researcher and future psychiatry resident, to get a glimpse at this field since my PhD was mainly experimental work. My main interest was about neuroimaging analysis, especially fMRI.

General remarks

It has been a very interesting week, although quite intense. I ended up being much tired, but with a feeling of having had a broad introduction to the many subfields of computational psychiatry. The program was well-organised and didactic, starting from the various clinical questions that need answering to how they are being worked on by researchers, with many days dedicated to introducing the various techniques one can use (see below).
The speakers were, for most of them, very didactic and managed to give comprehensive talks on complicated subjects. Nevertheless, there were only introductory courses ; do not expect to get out of Zurich and become an expert on Predictive Coding or Causal Modelling. You will have a broad understanding of the field, enough to know what you need to focus one for your research question. One strong advantage is the reading list at the end of nearly each course, which will ease one’s future work once one chose a specific technique.
I highly recommend taking the course onsite, although Zurich is known to be the most expensive European city, as meeting other students and chatting with the speakers after the courses are quite interesting. Moreover, onsite tutorials on the last day are a great way to start getting hands-on experience.

Program

As previously stated, the program is very didactic. The course goes on for 5 days and starts with a broad overview of the various psychiatric diseases during the first day. Indeed, only a handful of students comes from a medical background. These first courses are very interesting to grasp what clinicians lack to better treat patients.
Days 2 to 4 are dedicated to modeling. From the basics to slightly more advanced courses on perception or machine learning, both the theoretical approaches (e.g. Markov Decision Process or Dynamical Causal Modeling) and their applications to cognition (e.g. Predictive Coding or Active Inference) are described.
The 5th day is more conference-like, with great researchers (some of whom were speakers on the previous days) presenting their previous and ongoing work, building on what we learned during the week. It is very pleasant and interesting to understand how all the shenanigans we heard about can be applied to help patients.
You can find here the program of the course in 2024.
Hands-on tutorials are proposed on the 6th day, encompassing each topic studied during the course. It appears the quality of those are heterogenous but it is a good opportunity to start diving into computational psychiatry with people who can answer your firsts beginner’s questions.

Conclusion

10/10 would recommend. Indeed, it seems that a few students come back each year, probably to build up on their previous year’s work. I got back to Paris with clearer ideas about the field, a (huge) reading list to deepen my knowledge about neuroimaging analysis, and having met many incredible researchers in computational psychiatry!