Work

Expert-Guided Multimodal EEG–ECG Integration for Long-Term Neurological Outcome Prediction After Cardiac Arrest

Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Objectives To investigate whether integrating complementary cortical and cardiovascular neurophysiological biomarkers improves the characterization and internal prediction of neurological outcomes following cardiac arrest. Background Recovery after postanoxic coma involves cortical network organization, nonlinear cerebral dynamics, autonomic regulation and brain-heart communication. Multimodal neurophysiology offers a more comprehensive perspective than isolated measurements. Methods We analyzed 134 outcome-labeled post-cardiac-arrest coma records; 121 had complete predictors (48 favorable, 73 poor). The corresponding author (C.M.) performed expert visual EEG/ECG curation incorporating ICA-based artifact management. M2 comprised age, relative alpha/theta power and Hilbert-Huang complexity; M3 additionally incorporated frontal heartbeat-evoked responses (HER) and heart-rate variability (SDHR). Elastic Net, Gradient Boosting and within-fold algorithm selection were evaluated with repeated nested five-fold cross-validation. Results Gradient Boosting achieved an out-of-fold AUC of 0.738 for M3 versus 0.713 for M2 (difference 0.025; exploratory 95% CI -0.051 to 0.104). Nested selection yielded AUCs of 0.709 and 0.654, respectively (difference 0.055; 95% CI -0.022 to 0.135). Both incremental intervals included zero. Conclusions Expert-guided multimodal EEG-ECG integration offers a complementary framework for evaluating cortical function and brain-heart interactions. Internal discrimination was moderate; added predictive value of HER and HRV remains uncertain and requires independent validation.
Language: English
Field: Medicine
Open Access status: green
OpenAlex ID: W7236196483
Cites: 35
Cited by:
Related to: 10