Combining early post-resuscitation EEG and HRV features improves the prognostic performance in cardiac arrest model of rats

Explainable machine learning model based on EEG, ECG, and clinical features for predicting neurological outcomes in cardiac arrest patient

Early and accurate prediction of neurological outcomes in comatose patients following cardiac arrest is critical for informed clinical decision-making. Existing studies have predominantly focused on EEG for assessing brain injury, with some exploring ECG data. However, the integration of EEG, ECG, and clinical features remains insufficiently investigated, and its potential to enhance predictive accuracy has not been fully established. Moreover, the limited interpretability of current models poses significant barriers to clinical application. Using the I-CARE database, we analyzed EEG, ECG, and clinical data from comatose cardiac arrest patients. After rigorous preprocessing and feature engineering, machine learning models (Logistic Regression, SVM, Random Forest, and Gradient Boosting) were developed. Performance was evaluated through AUC-ROC, accuracy, sensitivity, and specificity, with SHAP applied to interpret feature contributions. Our multi-modal model outperformed single-modality models, achieving AUC values from 0.75 to 1.0. Notably, the model's accuracy peaked at a critical point within the 12-24 h window (e.g., 18 h, AUC = 1.0), surpassing EEG-only (AUC 0.7-0.8) and ECG-only (AUC < 0.6) models. SHAP identified Shockable Rhythm as the most influential feature (mean SHAP value 0.17), emphasizing its role in predictive accuracy. This study presents a novel multi-modal approach that significantly enhances early neurological outcome prediction in critical care. SHAP-based interpretability further supports clinical applicability, paving the way for more personalized patient management post-cardiac arrest.

COMBINATION OF HYPEROXYGENATION AND TARGETED TEMPERATURE MANAGEMENT IMPROVES FUNCTIONAL OUTCOMES OF POST CARDIAC ARREST SYNDROME IRRESPECTIVE OF CAUSES OF ARREST IN RATS

ABSTRACT

Background: The high mortality rates of patients who are resuscitated from cardiac arrest (CA) are attributed to post cardiac arrest syndrome (PCAS). This study evaluated the effect of hyperoxygenation and targeted temperature management (TTM) on PCAS in rats with different causes of CA. Methods and Results: One hundred sixty-eight Sprague-Dawley rats were equally divided into asphyxial and dysrhythmic groups. Animals were further randomized into four subgroups immediately after resuscitation: normoxia-normothermia (NO-NT), ventilated with 21% oxygen under normothermia; hyperoxia-normothermia (HO-NT), ventilated with 100% oxygen for 3 hours under normothermia; normoxia-hypothermia (NO-HT), ventilated with 21% oxygen for 3 hours under hypothermia; and hyperoxia-hypothermia (HO-HT), ventilated with 100% oxygen for 3 hours under hypothermia. Post resuscitation cardiac dysfunction, neurological recovery, and pathological analysis were assessed. For asphyxial CA, HO-NT and HO-HT (68.8% and 75.0%) had significantly higher survival than NO-NT and NO-HT (31.3% and 31.3%). For dysrhythmic CA, NO-HT and HO-HT (81.3% and 87.5%) had significantly higher survival than NO-NT and HO-NT (44.0% and 50.0%). When all of the rats were considered, the survival rate was much higher in HO-HT (81.3%). Compared with NO-NT (57.7% ± 14.9% and 40.3% ± 7.8%), the collagen volume fraction and the proportion of fluoro-jade B-positive area in HO-HT (14.0% ± 5.7% and 28.0% ± 13.3%) were significantly reduced. Conclusion: The beneficial effects of hyperoxygenation and TTM are dependent on the cause of arrest: hyperoxygenation benefits asphyxial, whereas TTM benefits dysrhythmic CA. The combination of hyperoxygenation and TTM could effectively improve the functional outcome of PCAS regardless of the cause of CA.

Deep and Wide Transfer Learning with Kernel Matching for Pooling Data from Electroencephalography and Psychological Questionnaires

Motor imagery (MI) promotes motor learning and encourages brain–computer interface systems that entail electroencephalogram (EEG) decoding. However, a long period of training is required to master brain rhythms’ self-regulation, resulting in users with MI inefficiency. We introduce a parameter-based approach of cross-subject transfer-learning to improve the performances of poor-performing individuals in MI-based BCI systems, pooling data from labeled EEG measurements and psychological questionnaires via kernel-embedding. To this end, a Deep and Wide neural network for MI classification is implemented to pre-train the network from the source domain. Then, the parameter layers are transferred to initialize the target network within a fine-tuning procedure to recompute the Multilayer Perceptron-based accuracy. To perform data-fusion combining categorical features with the real-valued features, we implement stepwise kernel-matching via Gaussian-embedding. Finally, the paired source–target sets are selected for evaluation purposes according to the inefficiency-based clustering by subjects to consider their influence on BCI motor skills, exploring two choosing strategies of the best-performing subjects (source space): single-subject and multiple-subjects. Validation results achieved for discriminant MI tasks demonstrate that the introduced Deep and Wide neural network presents competitive performance of accuracy even after the inclusion of questionnaire data.

Hyperoxygenation With Cardiopulmonary Resuscitation and Targeted Temperature Management Improves Post-Cardiac Arrest Outcomes in Rats

Background Oxygen plays a pivotal role in cardiopulmonary resuscitation (CPR) and postresuscitation intervention for cardiac arrest. However, the optimal method to reoxygenate patients has not been determined. This study investigated the effect of timing of hyperoxygenation on neurological outcomes in cardiac arrest/CPR rats treated with targeted temperature management. Methods and Results After induction of ventricular fibrillation, male Sprague-Dawley rats were randomized into 4 groups (n=16/group): (1) normoxic control; (2) O₂_CPR, ventilated with 100% O₂ during CPR; (3) O₂_CPR+postresuscitation, ventilated with 100% O₂ during CPR and the first 3 hours of postresuscitation; and (4) O₂_postresuscitation, ventilated with 100% O₂ during the first 3 hours of postresuscitation. Targeted temperature management was induced immediately after resuscitation and maintained for 3 hours in all animals. Postresuscitation hemodynamics, neurological recovery, and pathological analysis were assessed. Brain tissues of additional rats undergoing the same experimental procedure were harvested for ELISA-based quantification assays of oxidative stress-related biomarkers and compared with the sham-operated rats (n=6/group). We found that postresuscitation mean arterial pressure and quantitative electroencephalogram activity were significantly increased, whereas astroglial protein S100B, degenerated neurons, oxidative stress-related biomarkers, and neurologic deficit scores were significantly reduced in the O₂_CPR+postresuscitation group compared with the normoxic control group. In addition, 96-hour survival rates were significantly improved in all of the hyperoxygenation groups. Conclusions In this cardiac arrest/CPR rat model, hyperoxygenation coupled with targeted temperature management attenuates ischemia/reperfusion-induced injuries and improves survival rates. The beneficial effects of high-concentration oxygen are timing and duration dependent. Hyperoxygenation commenced with CPR, which improves outcomes when administered during hypothermia.

Inhaling Hydrogen Ameliorates Early Postresuscitation EEG Characteristics in an Asphyxial Cardiac Arrest Rat Model

Background

Electroencephalography (EEG) is commonly used to assess the neurological prognosis of comatose patients after cardiac arrest (CA). However, the early prognostic accuracy of EEG may be affected by postresuscitation interventions. Recent animal studies found that hydrogen inhalation after CA greatly improved neurological outcomes by selectively neutralizing highly reactive oxidants, but the effect of hydrogen inhalation on EEG recovery and its prognostication value are still unclear. The present study investigated the effects of hydrogen inhalation on early postresuscitation EEG characteristics in an asphyxial CA rat model.

Methods

Cardiopulmonary resuscitation was initiated after 5 min of untreated CA in 40 adult female Sprague-Dawley rats. Animals were randomized for ventilation with 98% oxygen plus 2% hydrogen (H2) or 98% oxygen plus 2% nitrogen (Ctrl) under normothermia for 1 h. EEG characteristics were continuously recorded for 4 h, and the relationships between quantitative EEG characteristics and 96 h neurological outcomes were investigated.

Results

No differences in baseline and resuscitation data were observed between groups, but the survival rate was significantly higher in the H2 group than in the Ctrl group (90% vs. 40%, P < 0.01). Compared to the Ctrl group, the H2 group showed a shorter burst onset time (21.85 [20.00-23.38] vs. 25.70 [22.48-30.05], P < 0.01) and time to normal trace (169.83 [161.63-208.55] vs. 208.39 [186.29-248.80], P < 0.01). Additionally, the burst suppression ratio (0.66 ± 0.09 vs. 0.52 ± 0.17, P < 0.01) and weighted-permutation entropy (0.47 ± 0.16 vs. 0.34 ± 0.13, P < 0.01) were markedly higher in the H2 group. The areas under the receiver operating characteristic curves for the 4 EEG characteristics in predicting survival were 0.82, 0.84, 0.88, and 0.83, respectively.

Conclusions

In this asphyxial CA rat model, the improved postresuscitation EEG characteristics for animals treated with hydrogen are correlated with the better 96 h neurological outcome and predicted survival.