EEG functional connectivity contributes to outcome prediction of postanoxic coma. Issue 6 (June 2021)

• Record Type: Journal Article
• Authors: Carrasco-Gómez, Martín; Keijzer, Hanneke M.; Ruijter, Barry J.; Bruña, Ricardo; Tjepkema-Cloostermans, Marleen C.; Hofmeijer, Jeannette; van Putten, Michel J.A.M.
• Is Part Of: Clinical neurophysiology. Volume 132:Issue 6(2021)
• Page Start: 1312
• Page End: 1320
• Publication Date: 2021-06
• Subjects: EEG functional connectivity -- Machine learning -- Postanoxic coma -- Intensive care -- Outcome prediction; EEG electroencephalography -- ICU intensive care unit -- CPC cerebral performance category -- SSEP somatosensory evoked potential -- COH coherence -- ciCOH corrected imaginary coherence -- PLV phase locking value -- ciPLV corrected imaginary phase locking value -- MI mutual information -- BT bagged tree -- LSVM linear support vector machine -- CI confidence interval, ROC, receiving operating characteristic -- AUC area under the curve; Neurophysiology -- Periodicals; Electroencephalography -- Periodicals; Electromyography -- Periodicals; Neurology -- Periodicals; 612.8
• DOI: 10.1016/j.clinph.2021.02.011
• Languages: English
• ISSNs: 1388-2457

Abstract

Highlights: Early EEG recordings under 48 h after cardiac arrest allow for prediction of outcome of coma patients. EEG-based functional connectivity features hold potential to improve outcome prediction of comatose patients after cardiac arrest. The most accurate prediction model combined functional connectivity and non-coupling EEG metrics, showing the best results to date. Abstract: Objective: To investigate the additional value of EEG functional connectivity features, in addition to non-coupling EEG features, for outcome prediction of comatose patients after cardiac arrest. Methods: Prospective, multicenter cohort study. Coherence, phase locking value, and mutual information were calculated in 19-channel EEGs at 12 h, 24 h and 48 h after cardiac arrest. Three sets of machine learning classification models were trained and validated with functional connectivity, EEG non-coupling features, and a combination of these. Neurological outcome was assessed at six months and categorized as "good" (Cerebral Performance Category [CPC] 1–2) or "poor" (CPC 3–5). Results: We included 594 patients (46% good outcome). A sensitivity of 51% (95% CI: 34–56%) at 100% specificity in predicting poor outcome was achieved by the best functional connectivity-based classifier at 12 h after cardiac arrest, while the best non-coupling-based model reached a sensitivity of 32% (0–54%) at 100% specificity using data at 12 h and 48 h. Combination of both sets of features achieved a sensitivity of 73% (50–77%) at 100% specificity. Conclusion: Functional connectivity measures improve EEG based prediction models for poor outcome of postanoxic coma. Significance: Functional connectivity features derived from early EEG hold potential to improve outcome prediction of coma after cardiac arrest.