Multimodal prediction of residual consciousness in the intensive care unit: the CONNECT-ME study
Moshgan Amiri,Patrick M Fisher,Federico Raimondo,Annette Sidaros,Melita Cacic Hribljan,Marwan H Othman,Ivan Zibrandtsen,Simon A Albrechtsen,Ove Bergdal,Adam Espe Hansen,Christian Hassager,Joan Lilja S Højgaard,Elisabeth Waldemar Jakobsen,Helene Ravnholt Jensen,Jacob Møller,Vardan Nersesjan,Miki Nikolic,Markus Harboe Olsen,Sigurdur Thor Sigurdsson,Jacobo D Sitt,Christine Sølling,Karen Lise Welling,Lisette M Willumsen,John Hauerberg,Vibeke Andrée Larsen,Martin Ejler Fabricius,Gitte Moos Knudsen,Jesper Kjærgaard,Kirsten Møller,Daniel Kondziella
DOI: https://doi.org/10.1093/brain/awac335
IF: 14.5
2022-09-14
Brain
Abstract:Functional MRI (fMRI) and EEG may reveal residual consciousness in patients with disorders of consciousness (DoC), as reflected by a rapidly expanding literature on chronic DoC. However, acute DoC is rarely investigated, although identifying residual consciousness is key to clinical decision-making in the intensive care unit (ICU). Therefore, the objective of the prospective, observational, tertiary center cohort, diagnostic phase IIb study 'Consciousness in neurocritical care cohort study using EEG and fMRI' (CONNECT-ME, NCT02644265) was to assess the accuracy of fMRI and EEG to identify residual consciousness in acute DoC in the ICU. Between April 2016 and November 2020, 87 acute DoC-patients with traumatic or non-traumatic brain injury were examined with repeated clinical assessments, fMRI and EEG. Resting-state EEG and EEG with external stimulations were evaluated by visual analysis, spectral band analysis and a Support Vector Machine (SVM) consciousness classifier. In addition, within- and between-network resting-state connectivity for canonical resting-state fMRI networks were assessed. Next, we used EEG and fMRI data at study enrollment in two different machine-learning algorithms (Random Forest and SVM with a linear kernel), to distinguish patients in a minimally conscious state or better (≥MCS) from those in coma or unresponsive wakefulness state (≤UWS), at time of study enrollment and at ICU-discharge (or before death). Prediction performances were assessed with area under the curve (AUC). Of 87 DoC-patients (mean age, 50.0 ± 18 years, 43% women), 51 (59%) were ≤ UWS and 36 (41%) were ≥ MCS at study enrollment. Thirty-one (36%) patients died in the ICU, including 28 who had life-sustaining therapy withdrawn. EEG and fMRI predicted consciousness levels at study enrollment and ICU-discharge, with maximum AUCs of 0.79 (95% CI 0.77-0.80) and 0.71 (95% CI 0.77-0.80), respectively. Models based on combined EEG and fMRI features predicted consciousness levels at study enrollment and ICU-discharge with maximum AUCs of 0.78 (95% CI 0.71-0.86) and 0.83 (95% CI 0.75-0.89), respectively, with improved positive predictive value and sensitivity. Overall, both machine-learning algorithms (SVM and Random Forest) performed equally well. In conclusion, we suggest that acute DoC prediction models in the ICU be based on a combination of fMRI and EEG features, regardless of the machine-learning algorithm used.
neurosciences,clinical neurology