Abstract
<title>Abstract</title> <p>Background Early neurological prognostication after cardiac arrest remains challenging despite advances in neurocritical care monitoring. Although quantitative electroencephalography (EEG) and electrocardiography (ECG) provide complementary information regarding cortical, autonomic, and brain–heart physiology, these signals are typically interpreted independently. We evaluated whether integrating multiple neurophysiological biomarkers into a unified computational monitoring framework improves prediction of long-term neurological outcome. Methods This retrospective multicenter study included 134 comatose adult survivors of cardiac arrest from the publicly available I-CARE database. Quantitative EEG spectral measures, Hilbert–Huang Transform (HHT)-derived complexity metrics, heartbeat-evoked responses (HERs), heart rate variability (HRV), and EEG microstate features were extracted from synchronized EEG and ECG recordings obtained during the acute post-resuscitation period. Feature selection was performed using Random Forest analysis, followed by multivariable logistic regression. Individual physiological domains and progressively integrated multimodal models were compared using receiver operating characteristic (ROC) analysis. Results The most informative biomarkers originated from complementary physiological domains, including gamma relative power, HHT complexity, frontal HER amplitude, alpha relative power, theta relative power, and SDHR. Progressive integration of physiological modalities consistently improved model discrimination. The final multimodal monitoring framework achieved an area under the ROC curve of 0.739 (95% CI 0.643–0.827), outperforming all individual neurophysiological domains while maintaining biological interpretability. Conclusions Combining cortical electrophysiology, nonlinear EEG dynamics, autonomic regulation, and brain–heart interactions provides more accurate neurological outcome prediction than isolated physiological biomarkers. This multimodal neurophysiological monitoring framework represents a promising computational strategy for quantitative bedside monitoring and clinical decision support after cardiac arrest.</p>