Subproject 7

Anterior cingulate cortex-based biomarker development for cognitive-behavioral therapy (CBT) response prediction in internalizing disorders

Tailoring treatment for mental disorders to individual patient characteristics is the aim of precision psychiatry and psychotherapy. This approach requires powerful predictors for the outcome of a given treatment on a specific patient. Previous research demonstrated that emotion regulation is central to both psychopathology and cognitive-behavior therapy (CBT) and that the anterior cingulate cortex (ACC) is a crucial hub region for emotion regulation by its integration in several core brain networks. We propose that a dedicated systems-driven perspective will enable us to develop neuroimaging-derived biomarkers capable of predicting (non-)response to CBT for internalizing disorders. In particular, our subproject will optimize ACC predictor information by employing three sophisticated methods: 1) an efficient gradient of brain connectivity encoding alterations in the neural basis of explicit-controlled and implicit-automatic emotion regulation, 2) a graph-theoretical characterization of ACC networks, and 3) a Bayesian hierarchical modeling approach to networks focusing on the ACC hub. We will train and fine-tune our predictive models on three retrospective datasets used within the Research Unit, and then test their prediction accuracies both retrospectively and also on the prospective, independent dataset collected by SP1. Together, this approach allows us to test the core hypothesis that a sophisticated ensemble learning strategy based on several metrics for ACC network integration will achieve high accuracy for predicting CBT non-response. Additionally, results from this project will be returned to the Research Unit, by providing an optimal composition of ACC-centered system characteristics to SP2, by comparing our predictive performance with other neuroimaging-based predictors from SP8 and SP9, and by investigating whether our predictive model is associated with markers based on emotion-regulation strategies and psychophysiological markers of emotion regulation from SP4 and SP5. Overall, by applying methodological improvements and a clear systems perspective on the ACC, we aim to construct a predictor of CBT non-response that will demonstrate powerful prediction performance while at the same time providing high reliability and good predictor interpretability.


Till Adam

Student Assistant

Rebecca Delfendahl, B.Sc.

Student Assistant

Dr. Kevin Hilbert

Principal Investigator

Charlotte Meinke, M.Sc.

Research Associate

Prof. Dr. Dr. Henrik Walter

Co-Spokesperson, Principal Investigator



Chavanne, A. V., Meinke, C., Langhammer, T., Roesmann, K., Böhnlein, J., Gathmann, B., Herrmann, M. J., Junghöfer, M., Klahn, A. L., Schwarzmeier, H., Seeger, F., Siminski, N., Straube, T., Dannlowski, U., Lueken, U., Leehr, E. J. & Hilbert, K. (2023). Individual-Level Prediction of exposure therapy outcome using structural and functional MRI data in Spider Phobia: a Machine-Learning Study. Depression and Anxiety, 2023, 1–11.


Meinke, C., Lueken, U., Walter, H., & Hilbert, K. (2023, May 10-13). Können Resting-State fMRT Konnektivität und maschinelles Lernen zur Prädiktion des individuellen Behandlungserfolgs genutzt werden? - Ein systematisches Review [Poster presentation]. 2. Deutscher Psychotherapie Kongress, Berlin, Germany.