Between patterns and predictions: Interpretable latent EEG representations for clinical insights.
4 affiliations
- 1Section on Computational Neurology, Department of Neurology and Berlin Institute of Health, Charité - Universitätsmedizin Berlin, Berlin, Germany; Institute for Theoretical Biology, Department of Biology, Humboldt-Universität zu Berlin, Berlin, Germany.
- 2Section on Computational Neurology, Department of Neurology and Berlin Institute of Health, Charité - Universitätsmedizin Berlin, Berlin, Germany.
- 3Harvard Medical School, Boston, MA, USA; Beth Israel Deaconess Medical Center, Boston, MA, USA.
- 4Section on Computational Neurology, Department of Neurology and Berlin Institute of Health, Charité - Universitätsmedizin Berlin, Berlin, Germany. Electronic address: [email protected].
Abstract
Electroencephalography (EEG) captures rich brain dynamics, yet in clinical practice this complexity is often reduced to simplified summaries or categorical labels, limiting its interpretability for decision-making. We tested the hypothesis that a pretrained latent embedding framework, the Universal Map of EEG (UM-EEG), can preserve clinically meaningful structure across heterogeneous datasets and provide a generalizable representation of brain states. We applied UM-EEG, without retraining, to three independent cohorts spanning distinct clinical contexts: long-term EEG recordings from cardiac arrest patients (n = 575), subarachnoid hemorrhage (n = 100), and routine clinical EEG recordings containing physiological and pathological patterns (n = 141). EEG segments were projected into a 128-dimensional space anchored by expert-derived reference states, including wakefulness, sleep stages, ictal-interictal continuum activity, and burst suppression. Across datasets, favorable outcome or physiological recordings were consistently located closer to healthy reference states, whereas poor outcome and pathological recordings shifted toward pathological regions of the embedding space. Trajectory-derived geometric and temporal features discriminated outcome in cardiac arrest (ROC-AUC 0.83) and subarachnoid hemorrhage (ROC-AUC 0.76), and distinguished physiological from pathological routine EEGs (ROC-AUC 0.93). In routine EEG, similarity relationships derived from embedding trajectories correlated with those derived from structured clinical reports, indicating that the latent space recapitulates clinically relevant organization. These findings show that a fixed, semantically structured EEG embedding generalizes across etiologies and recording settings, enabling prognostic stratification and contextual interpretation while preserving the relational structure of brain states.
Related
- Clinical characteristics, treatment patterns and long-term outcomes of warm and mixed autoimmune hemolytic anemia: insights from a real-world study
- FloREN: Decoding Immune Regulatory Networks through Interpretable Graph Transformer Patient Representations
- Learning Interpretable Tumor Microenvironment Representations by Fitting Pan-Cancer Cell State-Niche Correlation
- Pretrained gene representations transfer mean expression more broadly than spatial patterns in virtual spatial transcriptomics
- CortexLens: Systematically Decoding Interpretable Visual Patterns from Human Visual Cortex
BibTeX
@article{krumm2026between,
title = {{Between patterns and predictions: Interpretable latent EEG representations for clinical insights.}},
author = {Laura Krumm and Robert Terziev and Dominik D Kranz and M Brandon Westover and Christian Meisel},
journal = {NeuroImage. Clinical},
year = {2026},
month = oct,
doi = {10.1016/j.nicl.2026.104077},
url = {https://europepmc.org/article/MED/42887172},
note = {CC BY},
}
RIS
TY - JOUR
TI - Between patterns and predictions: Interpretable latent EEG representations for clinical insights.
AU - Laura Krumm
AU - Robert Terziev
AU - Dominik D Kranz
AU - M Brandon Westover
AU - Christian Meisel
PY - 2026
DA - 2026/10/14
JO - NeuroImage. Clinical
DO - 10.1016/j.nicl.2026.104077
UR - https://europepmc.org/article/MED/42887172
AB - Electroencephalography (EEG) captures rich brain dynamics, yet in clinical practice this complexity is often reduced to simplified summaries or categorical labels, limiting its interpretability for decision-making. We tested the hypothesis that a pretrained latent embedding framework, the Universal Map of EEG (UM-EEG), can preserve clinically meaningful structure across heterogeneous datasets and provide a generalizable representation of brain states. We applied UM-EEG, without retraining, to three independent cohorts spanning distinct clinical contexts: long-term EEG recordings from cardiac arrest patients (n = 575), subarachnoid hemorrhage (n = 100), and routine clinical EEG recordings containing physiological and pathological patterns (n = 141). EEG segments were projected into a 128-dimensional space anchored by expert-derived reference states, including wakefulness, sleep stages, ictal-interictal continuum activity, and burst suppression. Across datasets, favorable outcome or physiological recordings were consistently located closer to healthy reference states, whereas poor outcome and pathological recordings shifted toward pathological regions of the embedding space. Trajectory-derived geometric and temporal features discriminated outcome in cardiac arrest (ROC-AUC 0.83) and subarachnoid hemorrhage (ROC-AUC 0.76), and distinguished physiological from pathological routine EEGs (ROC-AUC 0.93). In routine EEG, similarity relationships derived from embedding trajectories correlated with those derived from structured clinical reports, indicating that the latent space recapitulates clinically relevant organization. These findings show that a fixed, semantically structured EEG embedding generalizes across etiologies and recording settings, enabling prognostic stratification and contextual interpretation while preserving the relational structure of brain states.
SN - 2213-1582
ER -