Promise and Limits of Hierarchical Dynamical RNNs for Individualized Resting-State fMRI
Carlotta Barkhau, Keyvan Mahjoory, M. Brenner, Ramona Leenings, C. Pellengahr, L. Fisch, N.R. Winter, Maximilian Konowski, Hamidreza Jamalabadi, Tilo T. J. Kircher, M. Junghoefer, U. Dannlowski, T. Hahn
This paper evaluates hierarchical shallow piecewise-linear RNNs for modeling individual resting-state fMRI, finding they capture functional connectivity but have limited predictive power for phenotypic variables.
Neuroscience Applied · Jan 1, 2026Peer-reviewedNeuroscience