Computational Psychiatry
Mental disorders, including anxiety and depression, contribute strongly to global disability, affecting around 1 in 8 of the global population and incurring significant health costs. For effective mental healthcare, patients must get the intervention that is best for them. Current treatment is only 30% effective in initial care, with patients sometimes going through multiple rounds of interventions, as it has proven hard to predict treatment success.
-
dr. Marijn van Wingerden
Principal Investigator
The promise of precision psychiatry is that computational approaches can be enlisted to predict the best individual treatment and thus deliver better, individualized care.
In the Computational Psychiatry lab, we use machine- and deep learning models to analyze disease progression and treatment response using demographic, clinical and biomarker data. These data include patient scores on clinical diagnostic tools, ecological momentary assessment, behavioral assessment such as reward sensitivity and non-invasive brain recordings including EEG recordings. With these multimodal, longitudinal data patterns we train our computational models to predict treatment success for a specific patient-treatment pairing. Such a personalized intervention should have a higher chance of success, reducing psychiatric disease burden for patients and society.
Projects
- Transdiagnostic Clustering of Psychiatric symptoms in a population sample (with Erasmus MC)
- BiOmarkers fOr perSonalizing Treatment of depression (BOOST)
- Predicting in-hospital deterioration (with ETZ)
Partners
- Psychiatric epidemiology, Erasmus MC
- Elisabeth-Tweesteden Hospital Tilburg
- Esculine Business Intelligence & Health Analytics
- WARN-D: building an early warning system for depression (PI: dr. Eiko Fried, Leiden University)