Assessing Brain Dynamics for Predicting Postanoxic Coma Recovery: An EEG Based Approach

Research output: Contribution to book/Conference proceedings/Anthology/ReportConference contributionContributedpeer-review

Abstract

Postanoxic coma is caused by global anoxia of the brain, most often due to cardiac arrest. Electroen-cephalography (EEG) has been shown to provide prognostic information, where synchronous EEG activity is linked to cortico-cortical connectivity and arousals are linked to cortico-subcortical connectivity. Previous studies indicated changes in both connectivity dimensions in coma patients. As part of the PhysioNet Challenge 2023, we (ibmt-PeakyFinders) investigated a novel approach to predict the recovery from cardiac arrest by evaluating brain dynamics. We used the time delay stability method to assess the coupling behavior between different EEG channels reflecting cortico-cortical connectivity and arousal detection for the assessment of cortico-subcortical connectivity. To monitor the development of brain activity over time, a feature vector was generated from different time steps and extended with patient metadata to predict the recovery outcome of postanoxic coma patients. By reaching a challenge score (CS) of 0.34 our team ranked place 30. Reduction to selected connectivity features increased the CS on a held-out subset of the training set by 52.6 %, but not on the hidden validation set. Our results indicate that selected connectivity features contain information to predict the outcome of recovery from postanoxic coma.

Details

Original languageEnglish
Title of host publication2023 Computing in Cardiology (CinC)
PublisherIEEE
Number of pages4
ISBN (electronic)979-8-3503-8252-5
Publication statusPublished - 26 Dec 2023
Peer-reviewedYes

Publication series

SeriesComputing in Cardiology
Volume50
ISSN2325-887X

External IDs

Scopus 85182329339
ORCID /0000-0001-6754-5257/work/165451288
ORCID /0000-0003-4012-0608/work/165451918
ORCID /0000-0003-2214-6505/work/165454585
ORCID /0000-0002-1984-580X/work/165454716
ORCID /0000-0003-0095-8051/work/176862288