O-46 EEG connectivity measures in prognostication of postanoxic coma patients. Issue 7 (July 2019)

• Record Type: Journal Article
• Authors: Gómez, Martín Carrasco; Keijzer, Hanneke; Hofmeijer, Jeannette; Ruijter, Barry J.; Fernández, Ricardo Bruña; Tjepkema-Cloostermans, Marleen C.; van Putten, Michel J.A.M
• Is Part Of: Clinical neurophysiology. Volume 130:Issue 7(2019:Jul.)
• Page Start: e36
• Publication Date: 2019-07
• Subjects: Neurophysiology -- Periodicals; Electroencephalography -- Periodicals; Electromyography -- Periodicals; Neurology -- Periodicals; 612.8
• DOI: 10.1016/j.clinph.2019.04.361
• Languages: English
• ISSNs: 1388-2457

Abstract

Abstract : Early prognostication in postanoxic comatose patients remains a challenge. In this work, we investigate EEG functional connectivity (FC) measures as a contributor to prediction of outcome in these patients, as well as their additional value when combined with other validated qEEG measures. A retrospective study of prospectively collected data was performed. Outcome was assessed at 6 months and categorized as "good" (CPC 1–2) or "poor" (CPC 3–5). Coherence, corrected imaginary coherence, phase locking value, corrected imaginary phase locking value, and mutual information were calculated in EEGs recorded at 12, 24, and 48 h after cardiac arrest (CA). Sets of machine learning classification models were trained and validated with 3 different grops of parameters: FC metrics, an already validated set of conventional qEEG measures, and a combination of both. We included 594 patients, with 46% presenting good outcome. A sensitivity of 50.8% (48–52% CI) in predicting poor outcome was achieved by the best FC-based classifier at 12 h after CA, while the best conventional qEEG-based model presented a sensitivity of 32.6% (28–37% CI) using data from 12 and 48 h after CA. The best combined model achieved a sensitivity of 69.3% (64–74% CI) with the 12 and 48 h dataset. All sensitivities for poor outcome prediction are at 100% specificity (98–100% CI). FC parameters derived from EEG contribute to outcome prediction of postanoxic coma by themselves and are complementary to a validated model of classic EEG parameters, showing better performance than methods currently used in clinical practise and newly proposed.