Early electroencephalography (EEG) contains valuable information for predicting neurological recovery in comatose patients after cardiac arrest. As part of the George B. Moody PhysioNet Challenge 2023, our team, TUD_EEG, developed a novel ensembling approach that combines two pipelines with different directions of information transfer between patient-level and segment-level descriptions. Using both EEG and patient clinical information, our model achieved a Challenge score of 0.72 (3rd place out of 34 eligible teams) on the hidden test set.