Quantitative EEG predicts outcomes in children after cardiac arrest

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.

Prediction of Survival After Pediatric Cardiac Arrest Using Quantitative EEG and Machine Learning Techniques

BACKGROUND AND OBJECTIVES

Early neuroprognostication in children with reduced consciousness after cardiac arrest (CA) is a major clinical challenge. EEG is frequently used for neuroprognostication in adults, but has not been sufficiently validated for this indication in children. Using machine learning techniques, we studied the predictive value of quantitative EEG (qEEG) features for survival 12 months after CA, based on EEG recordings obtained 24 hours after CA in children. The results were confirmed through visual analysis of EEG background patterns.

METHODS

This is a retrospective single-center study including children (0-17 years) with CA, who were subsequently admitted to the pediatric intensive care unit (PICU) of a tertiary care hospital between 2012 and 2021 after return of circulation (ROC) and were monitored using EEG at 24 hours after ROC. Signal features were extracted from a 30-minute EEG segment 24 hours after CA and used to train a random forest model. The background pattern from the same EEG fragment was visually classified. The primary outcome was survival or death 12 months after CA. Analysis of the prognostic accuracy of the model included calculation of receiver-operating characteristic and predictive values. Feature contribution to the model was analyzed using Shapley values.

RESULTS

Eighty-six children were included (in-hospital CA 27%, out-of-hospital CA 73%). The median age at CA was 2.6 years; 53 (62%) were male. Mortality at 12 months was 56%; main causes of death on the PICU were withdrawal of life-sustaining therapies because of poor neurologic prognosis (52%) and brain death (31%). The random forest model was able to predict death at 12 months with an accuracy of 0.77 and positive predictive value of 1.0. Continuity and amplitude of the EEG signal were the signal parameters most contributing to the model classification. Visual analysis showed that no patients with a background pattern other than continuous with amplitudes exceeding 20 μV were alive after 12 months.

DISCUSSION

Both qEEG and visual EEG background classification for registrations obtained 24 hours after ROC form a strong predictor of nonsurvival 12 months after CA in children.

Background EEG Suppression Ratio for Early Detection of Cerebral Injury in Pediatric Cardiac Arrest

BACKGROUND

Our objective was to assess the utility of the 1-h suppression ratio (SR) as a biomarker of cerebral injury and neurologic prognosis after cardiac arrest (CA) in the pediatric hospital setting.

METHODS

Prospectively, we reviewed data from children presenting after CA and monitored by continuous electroencephalography (cEEG). Patients aged 1 month to 21 years were included. The SR, a quantitative measure of low-voltage cEEG (≤ 3 µV) content, was dichotomized as present or absent if there was > 0% suppression for one continuous hour. A multivariate logistic regression analysis was performed including age, sex, type of CA (i.e., in-hospital or out-of-hospital), and the presence of SR as a predictor of global anoxic cerebral injury as confirmed by magnetic resonance imaging (MRI).

RESULTS

We included 84 patients with a median age of 4 years (interquartile range 0.9-13), 64% were male, and 49% (41/84) had in-hospital CA. Cerebral injury was seen in 50% of patients, of whom 65% had global injury. One-hour SR presence, independent of amount, predicted cerebral injury with 81% sensitivity (95% confidence interval (CI) (66-91%) and 98% specificity (95% CI 88-100%). Multivariate logistic regression analyses indicated that SR was a significant predictor of both cerebral injury (β = 6.28, p < 0.001) and mortality (β = 3.56, p < 0.001).

CONCLUSIONS

The SR a sensitive and specific marker of anoxic brain injury and post-CA mortality in the pediatric population. Once detected in the post-CA setting, the 1-h SR may be a useful threshold finding for deployment of early neuroprotective strategies prior or for prompting diagnostic neuroimaging.

Association of bilaterally suppressed EEG amplitudes and outcomes in critically ill children

Background and objectives

Amplitude-integrated EEG (aEEG) is used to assess electrocortical activity in pediatric intensive care if (continuous) full channel EEG is unavailable but evidence regarding the meaning of suppressed aEEG amplitudes in children remains limited. This retrospective cohort study investigated the association of suppressed aEEG amplitudes in critically ill children with death or decline of neurological functioning at hospital discharge.

Methods

Two hundred and thirty-five EEGs derived from individual patients <18 years in the pediatric intensive care unit at the University Hospital Essen (Germany) between 04/2014 and 07/2021, were converted into aEEGs and amplitudes analyzed with respect to age-specific percentiles. Crude and adjusted odds ratios (OR) for death, and functional decline at hospital discharge in patients with bilateral suppression of the upper or lower amplitude below the 10th percentile were calculated. Sensitivity, specificity, positive (PPV) and negative predictive values (NPV) were assessed.

Results

The median time from neurological insult to EEG recording was 2 days. PICU admission occurred due to neurological reasons in 43% and patients had high overall disease severity. Thirty-three (14%) patients died and 68 (29%) had a functional decline. Amplitude suppression was observed in 48% (upper amplitude) and 57% (lower amplitude), with unilateral suppression less frequent than bilateral suppression. Multivariable regression analyses yielded crude ORs between 4.61 and 14.29 and adjusted ORs between 2.55 and 8.87 for death and functional decline if upper or lower amplitudes were bilaterally suppressed. NPVs for bilaterally non-suppressed amplitudes were above 95% for death and above 83% for pediatric cerebral performance category Scale (PCPC) decline, whereas PPVs ranged between 22 and 32% for death and 49-52% for PCPC decline.

Discussion

This study found a high prevalence of suppressed aEEG amplitudes in critically ill children. Bilaterally normal amplitudes predicted good outcomes, whereas bilateral suppression was associated with increased odds for death and functional decline. aEEG assessment may serve as an element for risk stratification of PICU patients if conventional EEG is unavailable with excellent negative predictive abilities but requires additional information to identify patients at risk for poor outcomes.

EEG Pattern With Spectral Analysis Can Prognosticate Good and Poor Neurologic Outcomes After Cardiac Arrest

PURPOSE

To investigate the prognostic value of a simple stratification system of electroencephalographical (EEG) patterns and spectral types for patients after cardiac arrest.

METHODS

In this prospectively enrolled cohort, using manually selected EEG segments, patients after cardiac arrest were stratified into five independent EEG patterns (based on background continuity and burden of highly epileptiform discharges) and four independent power spectral types (based on the presence of frequency components). The primary outcome is cerebral performance category (CPC) at discharge. Results from multimodal prognostication testing were included for comparison.

RESULTS

Of a total of 72 patients, 6 had CPC 1-2 by discharge, all of whom had mostly continuous EEG background without highly epileptiform activity at day 3. However, for the same EEG background pattern at day 3, 19 patients were discharged at CPC 3 and 15 patients at CPC 4-5. After adding spectral analysis, overall sensitivity for predicting good outcomes (CPC 1-2) was 83.3% (95% confidence interval 35.9% to 99.6%) and specificity was 97.0% (89.5% to 99.6%). In this cohort, standard prognostication testing all yielded 100% specificity but low sensitivity, with imaging being the most sensitive at 54.1% (36.9% to 70.5%).

CONCLUSIONS

Adding spectral analysis to qualitative EEG analysis may further improve the diagnostic accuracy of EEG and may aid developing novel measures linked to good outcomes in postcardiac arrest coma.

Electroencephalogram-based machine learning models to predict neurologic outcome after cardiac arrest: A systematic review

AIM OF THE REVIEW

The primary aim of this systematic review was to investigate the most common electroencephalogram (EEG)-based machine learning (ML) model with the highest Area Under Receiver Operating Characteristic Curve (AUC) in two ML categories, conventional ML and Deep Neural Network (DNN), to predict the neurologic outcomes after cardiac arrest; the secondary aim was to investigate common EEG features applied to ML models.

METHODS

Systematic search of medical literature from PubMed and engineering literature from Compendex up to June 2, 2023. One reviewer screened studies that used EEG-based ML models to predict the neurologic outcomes after cardiac arrest. Four reviewers validated that the studies met selection criteria. Nine variables were manually extracted. The top-five common EEG features were calculated. We evaluated each study's risk of bias using the Quality in Prognosis Studies guideline.

RESULTS

Out of 351 identified studies, 17 studies met the inclusion criteria. Random Forest (RF) (n = 7) was the most common ML model in the conventional ML category (n = 11), followed by Convolutional Neural Network (CNN) (n = 4) in the DNN category (n = 6). The AUCs for RF ranged between 0.8 and 0.97, while CNN had AUCs between 0.7 and 0.92. The top-three commonly used EEG features were band power (n = 12), Shannon's Entropy (n = 11), burst-suppression ratio (n = 9).

CONCLUSIONS

RF and CNN were the two most common ML models with the highest AUCs for predicting the neurologic outcomes after cardiac arrest. Using a multimodal model that combines EEG features and electronic health record data may further improve prognostic performance.

Review of Noninvasive Neuromonitoring Modalities in Children II: EEG, qEEG

Critically ill children with acute neurologic dysfunction are at risk for a variety of complications that can be detected by noninvasive bedside neuromonitoring. Continuous electroencephalography (cEEG) is the most widely available and utilized form of neuromonitoring in the pediatric intensive care unit. In this article, we review the role of cEEG and the emerging role of quantitative EEG (qEEG) in this patient population. cEEG has long been established as the gold standard for detecting seizures in critically ill children and assessing treatment response, and its role in background assessment and neuroprognostication after brain injury is also discussed. We explore the emerging utility of both cEEG and qEEG as biomarkers of degree of cerebral dysfunction after specific injuries and their ability to detect both neurologic deterioration and improvement.

Initial study on quantitative electroencephalographic analysis of bioelectrical activity of the brain of children with fetal alcohol spectrum disorders (FASD) without epilepsy

Fetal alcohol spectrum disorders (FASD) are spectrum of neurodevelopmental conditions associated with prenatal alcohol exposure. The FASD manifests mostly with facial dysmorphism, prenatal and postnatal growth retardation, and selected birth defects (including central nervous system defects). Unrecognized and untreated FASD leads to severe disability in adulthood. The diagnosis of FASD is based on clinical criteria and neither biomarkers nor imaging tests can be used in order to confirm the diagnosis. The quantitative electroencephalography (QEEG) is a type of EEG analysis, which involves the use of mathematical algorithms, and which has brought new possibilities of EEG signal evaluation, among the other things-the analysis of a specific frequency band. The main objective of this study was to identify characteristic patterns in QEEG among individuals affected with FASD. This study was of a pilot prospective study character with experimental group consisting of patients with newly diagnosed FASD and of the control group consisting of children with gastroenterological issues. The EEG recordings of both groups were obtained, than analyzed using a commercial QEEG module. As a results we were able to establish the dominance of the alpha rhythm over the beta rhythm in FASD-participants compared to those from the control group, mostly in frontal and temporal regions. Second important finding is an increased theta/beta ratio among patients with FASD. These findings are consistent with the current knowledge on the pathological processes resulting from the prenatal alcohol exposure. The obtained results and conclusions were promising, however, further research is necessary (and planned) in order to validate the use of QEEG tools in FASD diagnostics.

Identifying biomarkers for epilepsy after cerebral malaria in Zambian children: rationale and design of a prospective observational study

INTRODUCTION

Malaria affecting the central nervous system (CM) is a major contributor to paediatric epilepsy in resource-poor settings, with 10%-16% of survivors developing epilepsy within 2 years of infection. Despite high risk for post-malaria epilepsy (PME), biomarkers indicating which CM survivors will develop epilepsy are absent. Such biomarkers are essential to identify those at highest risk who might benefit most from close surveillance and/or preventive treatments. Electroencephalography (EEG) contains signals (specifically gamma frequency activity), which are correlated with higher risk of PME and provide a biomarker for the development of epilepsy. We propose to study the sensitivity of quantitative and qualitative EEG metrics in predicting PME, and the potential increased sensitivity of this measure with additional clinical metrics. Our goal is to develop a predictive PME index composed of EEG and clinical history metrics that are highly feasible to obtain in low-resourced regions.

METHODS AND ANALYSES

This prospective observational study being conducted in Eastern Zambia will recruit 250 children aged 6 months to 11 years presenting with acute CM and follow them for two years. Children with pre-existing epilepsy diagnoses will be excluded. Outcome measures will include qualitative and quantitative analysis of routine EEG recordings, as well as clinical metrics in the acute and subacute period, including histidine-rich protein 2 levels of parasite burden, depth and length of coma, presence and severity of acute seizures, presence of hypoglycaemia, maximum temperature and 1-month post-CM neurodevelopmental assessment scores. We will test the performance of these EEG and clinical metrics in predicting development of epilepsy through multivariate logistic regression analyses.

ETHICS AND DISSEMINATION

This study has been approved by the Boston Children's Hospital Institutional Review Board, University of Zambia Biomedical Research Ethics Committee, and National Health Research Authority of Zambia. Results will be disseminated locally in Zambia followed by publication in international, open access, peer-reviewed journals when feasible.

Predicting Neurological Outcome From Electroencephalogram Dynamics in Comatose Patients After Cardiac Arrest With Deep Learning

OBJECTIVE

Most cardiac arrest patients who are successfully resuscitated are initially comatose due to hypoxic-ischemic brain injury. Quantitative electroencephalography (EEG) provides valuable prognostic information. However, prior approaches largely rely on snapshots of the EEG, without taking advantage of temporal information.

METHODS

We present a recurrent deep neural network with the goal of capturing temporal dynamics from longitudinal EEG data to predict long-term neurological outcomes. We utilized a large international dataset of continuous EEG recordings from 1,038 cardiac arrest patients from seven hospitals in Europe and the US. Poor outcome was defined as a Cerebral Performance Category (CPC) score of 3-5, and good outcome as CPC score 0-2 at 3 to 6-months after cardiac arrest. Model performance is evaluated using 5-fold cross validation.

RESULTS

The proposed approach provides predictions which improve over time, beginning from an area under the receiver operating characteristic curve (AUC-ROC) of 0.78 (95% CI: 0.72-0.81) at 12 hours, and reaching 0.88 (95% CI: 0.85-0.91) by 66 h after cardiac arrest. At 66 h, (sensitivity, specificity) points of interest on the ROC curve for predicting poor outcomes were (32,99)%, (55,95)%, and (62,90)%, (99,23)%, (95,47)%, and (90,62)%; whereas for predicting good outcome, the corresponding operating points were (17,99)%, (47,95)%, (62,90)%, (99,19)%, (95,48)%, (70,90)%. Moreover, the model provides predicted probabilities that closely match the observed frequencies of good and poor outcomes (calibration error 0.04).

CONCLUSIONS AND SIGNIFICANCE

These findings suggest that accounting for EEG trend information can substantially improve prediction of neurologic outcomes for patients with coma following cardiac arrest.

Multimodal Neurologic Monitoring in Children With Acute Brain Injury

Children with acute neurologic illness are at high risk of mortality and long-term neurologic disability. Severe traumatic brain injury, cardiac arrest, stroke, and central nervous system infection are often complicated by cerebral hypoxia, hypoperfusion, and edema, leading to secondary neurologic injury and worse outcome. Owing to the paucity of targeted neuroprotective therapies for these conditions, management emphasizes close physiologic monitoring and supportive care. In this review, we will discuss advanced neurologic monitoring strategies in pediatric acute neurologic illness, emphasizing the physiologic concepts underlying each tool. We will also highlight recent innovations including novel monitoring modalities, and the application of neurologic monitoring in critically ill patients at risk of developing neurologic sequelae.

State of the Art of Machine Learning-Enabled Clinical Decision Support in Intensive Care Units: Literature Review

Background

Modern clinical care in intensive care units is full of rich data, and machine learning has great potential to support clinical decision-making. The development of intelligent machine learning–based clinical decision support systems is facing great opportunities and challenges. Clinical decision support systems may directly help clinicians accurately diagnose, predict outcomes, identify risk events, or decide treatments at the point of care.

Objective

We aimed to review the research and application of machine learning–enabled clinical decision support studies in intensive care units to help clinicians, researchers, developers, and policy makers better understand the advantages and limitations of machine learning–supported diagnosis, outcome prediction, risk event identification, and intensive care unit point-of-care recommendations.

Methods

We searched papers published in the PubMed database between January 1980 and October 2020. We defined selection criteria to identify papers that focused on machine learning–enabled clinical decision support studies in intensive care units and reviewed the following aspects: research topics, study cohorts, machine learning models, analysis variables, and evaluation metrics.

Results

A total of 643 papers were collected, and using our selection criteria, 97 studies were found. Studies were categorized into 4 topics—monitoring, detection, and diagnosis (13/97, 13.4%), early identification of clinical events (32/97, 33.0%), outcome prediction and prognosis assessment (46/97, 47.6%), and treatment decision (6/97, 6.2%). Of the 97 papers, 82 (84.5%) studies used data from adult patients, 9 (9.3%) studies used data from pediatric patients, and 6 (6.2%) studies used data from neonates. We found that 65 (67.0%) studies used data from a single center, and 32 (33.0%) studies used a multicenter data set; 88 (90.7%) studies used supervised learning, 3 (3.1%) studies used unsupervised learning, and 6 (6.2%) studies used reinforcement learning. Clinical variable categories, starting with the most frequently used, were demographic (n=74), laboratory values (n=59), vital signs (n=55), scores (n=48), ventilation parameters (n=43), comorbidities (n=27), medications (n=18), outcome (n=14), fluid balance (n=13), nonmedicine therapy (n=10), symptoms (n=7), and medical history (n=4). The most frequently adopted evaluation metrics for clinical data modeling studies included area under the receiver operating characteristic curve (n=61), sensitivity (n=51), specificity (n=41), accuracy (n=29), and positive predictive value (n=23).

Conclusions

Early identification of clinical and outcome prediction and prognosis assessment contributed to approximately 80% of studies included in this review. Using new algorithms to solve intensive care unit clinical problems by developing reinforcement learning, active learning, and time-series analysis methods for clinical decision support will be greater development prospects in the future.

Early Neuroprognostication Using Frontal Spectrograms in Moderately Sedated Cardiac Arrest Patients

Introduction. The integrated suppression ratio throughout all electroencephalography (EEG) patterns has rarely been studied. The aim of this study was to evaluate the clinical utility of the suppression ratio and hyperactivity of EEG on spectrograms. Methods. This prospective observational study included 73 cardiac arrest patients. Hardwired frontal EEG monitoring with spectrograms (color density spectral arrays, CDSA) was used to predict neurological outcomes. The mean suppression ratio (MSR) and hyperactivity in the high-frequency band (HHF) in the spectrogram were investigated in moderately sedated patients. Sedative doses were considered to estimate the MSR, which was automatically measured. Results. Using propofol 30 to 40 µg/kg/min and remifentanil 0.1 to 0.15 µg/kg/min, all the patients with an MSR >30% died. At day 2, the MSR in patients with a good outcome was 0%. The cut off values were different as an MSR >30% at day 1 (AUC 0.815) and an MSR >1% at day 2 (AUC 0.891). Of the patients with an MSR ≤30%, HHF was the greatest predictor of a poor outcome (OR 12.858, P = .006). The best predictors of a poor outcome using the spectrogram were suppression ratio (SR) >30% or HHF at day 1 (AUC 0.88) and SR >1% or HHF at day 2 (AUC 0.909). Conclusions. The use of MSR and HHF in frontal spectrograms is convenient and may be successfully employed for early neuroprognostication in moderately sedated cardiac arrest patients. However, spectrograms should be used with electroencephalogram considering the effects of sedatives because of the imperfect detection of electrographic seizures and artifacts.

Brain Data in Pediatric Disorders of Consciousness: Special Considerations

SUMMARY

The diagnosis and management of disorders of consciousness in children continue to present a clinical, research, and ethical challenge. Though the practice guidelines for diagnosis and management of disorders of consciousness in adults are supported by decades of empirical and pragmatic evidence, similar guidelines for infants and children are lacking. The maturing conscious experience and the limited behavioral repertoire to report consciousness in this age group restrict extrapolation from the adult literature. Equally challenging is the process of heightened structural and functional neuroplasticity in the developing brain, which adds a layer of complexity to the investigation of the neural correlates of consciousness in infants and children. This review discusses the clinical assessment of pediatric disorders of consciousness and delineates the diagnostic and prognostic utility of neurophysiological and neuroimaging correlates of consciousness. The potential relevance of these correlates for the developing brain based on existing theoretical models of consciousness in adults is outlined.

Using EEG in Resource-Limited Areas: Comparing Qualitative and Quantitative Interpretation Methods in Cerebral Malaria

BACKGROUND

Our goal was to compare the strength of association and predictive ability of qualitative and quantitative electroencephalographic (EEG) factors with the outcomes of death and neurological disability in pediatric cerebral malaria (CM).

METHODS

We enrolled children with a clinical diagnosis of CM admitted to Queen Elizabeth Central Hospital (Blantyre, Malawi) between 2012 and 2017. A routine-length EEG was performed within four hours of admission. EEG data were independently interpreted using qualitative and quantitative methods by trained pediatric neurophysiologists. EEG interpreters were unaware of patient discharge outcome.

RESULTS

EEG tracings from 194 patients were reviewed. Multivariate modeling revealed several qualitative and quantitative EEG variables that were independently associated with outcomes. Quantitative methods modeled on mortality had better goodness of fit than qualitative ones. When modeled on neurological morbidity in survivors, goodness of fit was better for qualitative methods. When the probabilities of an adverse outcome were calculated using multivariate regression coefficients, only the model of quantitative EEG variables regressed on the neurological sequelae outcome showed clear separation between outcome groups.

CONCLUSIONS

Multiple qualitative and quantitative EEG factors are associated with outcomes in pediatric CM. It may be possible to use quantitative EEG factors to create automated methods of study interpretation that have similar predictive abilities for outcomes as human-based interpreters, a rare resource in many malaria-endemic areas. Our results provide a proof-of-concept starting point for the development of quantitative EEG interpretation and prediction methodologies useful in resource-limited settings.

Quantitative Electroencephalography for Early Detection of Elevated Intracranial Pressure in Critically Ill Children: Case Series and Proposed Protocol

OBJECTIVE

To describe quantitative EEG (electroencephalography) suppression ratio in children with increased intracranial pressure comparing acute suppression ratio changes to imaging and/or examination findings.

METHODS

We retrospectively reviewed the suppression ratio from patients with neuroimaging and /or examination findings of increased intracranial pressure while on continuous EEG. The time of the first change in the suppression ratio was compared to the time of the first image and/or examination change confirming increased intracranial pressure.

RESULTS

Thirteen patients with a median age of 3.1 years(interquartile range 1.8-6.3) had a rise in the suppression ratio with median time from identification to acute neuroimaging or examination of increased intracranial pressure of 3.12 hours (interquartile range 2.2-33.5) after the first increase in the suppression ratio.

CONCLUSIONS

Acute suppression ratio increase is seen prior to imaging and/or examination findings of increased intracranial pressure. With further study, the suppression ratio can be targeted with intracranial pressure-lowering agents to prevent morbidity and mortality associated with increased intracranial pressure.

Acute Neurologic Dysfunction in Critically Ill Children: The PODIUM Consensus Conference

CONTEXT

Acute neurologic dysfunction is common in critically ill children and contributes to outcomes and end of life decision-making.

OBJECTIVE

To develop consensus criteria for neurologic dysfunction in critically ill children by evaluating the evidence supporting such criteria and their association with outcomes.

DATA SOURCES

Electronic searches of PubMed and Embase were conducted from January 1992 to January 2020, by using a combination of medical subject heading terms and text words to define concepts of neurologic dysfunction, pediatric critical illness, and outcomes of interest.

STUDY SELECTION

Studies were included if the researchers evaluated critically ill children with neurologic injury, evaluated the performance characteristics of assessment and scoring tools to screen for neurologic dysfunction, and assessed outcomes related to mortality, functional status, organ-specific outcomes, or other patient-centered outcomes. Studies with an adult population or premature infants (≤36 weeks' gestational age), animal studies, reviews or commentaries, case series with sample size ≤10, and studies not published in English with an inability to determine eligibility criteria were excluded.

DATA EXTRACTION

Data were abstracted from each study meeting inclusion criteria into a standard data extraction form by task force members.

DATA SYNTHESIS

The systematic review supported the following criteria for neurologic dysfunction as any 1 of the following: (1) Glasgow Coma Scale score ≤8; (2) Glasgow Coma Scale motor score ≤4; (3) Cornell Assessment of Pediatric Delirium score ≥9; or (4) electroencephalography revealing attenuation, suppression, or electrographic seizures.

CONCLUSIONS

We present consensus criteria for neurologic dysfunction in critically ill children.

Predicting neurological outcome in comatose patients after cardiac arrest with multiscale deep neural networks

OBJECTIVE

Electroencephalography (EEG) is an important tool for neurological outcome prediction after cardiac arrest. However, the complexity of continuous EEG data limits timely and accurate interpretation by clinicians. We develop a deep neural network (DNN) model to leverage complex EEG trends for early and accurate assessment of cardiac arrest coma recovery likelihood.

METHODS

We developed a multiscale DNN combining convolutional neural networks (CNN) and recurrent neural networks (long short-term memory [LSTM]) using EEG and demographic information (age, gender, shockable rhythm) from a multicenter cohort of 1,038 cardiac arrest patients. The CNN learns EEG feature representations while the multiscale LSTM captures short-term and long-term EEG dynamics on multiple time scales. Poor outcome is defined as a Cerebral Performance Category (CPC) score of 3-5 and good outcome as CPC score 1-2 at 3-6 months after cardiac arrest. Performance is evaluated using area under the receiver operating characteristic curve (AUC) and calibration error.

RESULTS

Model performance increased with EEG duration, with AUC increasing from 0.83 (95% Confidence Interval [CI] 0.79-0.87 at 12h to 0.91 (95%CI 0.88-0.93) at 66h. Sensitivity of good and poor outcome prediction was 77% and 75% at a specificity of 90%, respectively. Sensitivity of poor outcome was 50% at a specificity of 99%. Predicted probability was well matched to the observation frequency of poor outcomes, with a calibration error of 0.11 [0.09-0.14].

CONCLUSIONS

These results demonstrate that incorporating EEG evolution over time improves the accuracy of neurologic outcome prediction for patients with coma after cardiac arrest.

Associations between initial serum pH value and outcomes of pediatric out-of-hospital cardiac arrest

INTRODUCTION

Pediatric out-of-hospital cardiac arrest (OHCA) is one of the most critical conditions seen in the emergency department (ED). Although initial serum pH value is reported to be associated with outcome in adult OHCA patients, the association is unclear in pediatric OHCA patients. Thus, we aimed to identify the association between initial pH value and outcome among pediatric OHCA patients.

METHODS

This study was a retrospective analysis of a multicenter prospective cohort registry (Japanese Association for Acute Medicine out-of-hospital cardiac arrest registry) from 87 hospitals in Japan. We included pediatric OHCA patients younger than 16 years of age who were registered in this registry between June 2014 and December 2017. Of the 34,754 patients in the database, 458 patients were ultimately included in the analysis. We equally divided the patients into four groups, based on their initial pH value, and conducted a multivariate logistic regression analysis to calculate the adjusted odds ratios of the initial pH value on hospital arrival with their 95% confidence intervals for the primary outcome.

RESULTS

The median (interquartile range) age was 1 (0-6) year, and 77.9% (357/458) of the first monitored rhythm was asystole. The primary outcome was 1-month survival. The overall 1-month survival was 13.3% (61/458), and a 1-month favorable neurologic outcome was seen in 5.2% (24/458) of cases. The adjusted odds ratios and 95% confidence intervals for the pH 6.81-6.64, pH 6.63-6.47, pH <6.47, and pH unknown groups compared with the pH ≥6.82 group for 1-month survival were 0.39 (0.16-0.97), 0.13 (0.04-0.44), 0.03 (0.00-0.24), and 0.07 (0.02-0.21), respectively.

CONCLUSIONS

This study demonstrated the association between the initial pH value on hospital arrival and 1-month survival among pediatric OHCA patients.

EEG markers predictive of epilepsy risk in pediatric cerebral malaria - A feasibility study

OBJECTIVE

Cerebral malaria (CM) affects 500,000 million children annually, 10% whom develop epilepsy within two years. Acute identification of biomarkers for post-CM epilepsy would allow for follow-up of the highest risk populations in resource-limited regions. We investigated the utility of electroencephalogram (EEG) and clinical metrics obtained during acute CM infection for predicting epilepsy.

METHODS

We analyzed 70 EEGs recorded within 24 h of admission for CM hospitalization obtained during the Blantyre Malaria Project Epilepsy Study (2005-2007), a prospective cohort study of pediatric CM survivors. While all studies underwent spectral analyses for comparisons of mean power band frequencies, a subset of EEGs from the 10 subjects who developed epilepsy and 10 age- and sex-matched controls underwent conventional visual analysis. Findings were tested for relationships to epilepsy outcomes.

RESULTS

Ten of the 70 subjects developed epilepsy. There were no significant differences between groups that were analyzed via visual EEG review; however, spectral EEG analyses revealed a significantly higher gamma-delta power ratio in CM survivors who developed epilepsy (0.23 ± 0.10) than in those who did not (0.16 ± 0.06), p = 0.003. Excluding potential confounders, multivariable logistic-regression analyses found relative gamma power (p = 0.003) and maximum temperature during admission (p = 0.03) significant and independent predictors of post-CM epilepsy, with area under receiver operating characteristics (AUROC) curve of 0.854.

CONCLUSIONS

We found that clinical and EEG metrics acquired during acute CM presentation confer risk of post-CM epilepsy. Further studies are required to investigate the utility of gamma activity as a potential biomarker of epileptogenesis and study this process over time. Additionally, resource limitations currently prevent follow-up of all CM cases to surveil for epilepsy, and identification of acute biomarkers in this population would offer the opportunity to allocate resources more efficiently.

IRIS: A Modular Platform for Continuous Monitoring and Caretaker Notification in the Intensive Care Unit

OBJECTIVE

New approaches are needed to interpret large amounts of physiologic data continuously recorded in the ICU. We developed and prospectively validated a versatile platform (IRIS) for real-time ICU physiologic monitoring, clinical decision making, and caretaker notification.

METHODS

IRIS was implemented in the neurointensive care unit to stream multimodal time series data, including EEG, intracranial pressure (ICP), and brain tissue oxygenation (PbtO2), from ICU monitors to an analysis server. IRIS was applied for 364 patients undergoing continuous EEG, 26 patients undergoing burst suppression monitoring, and four patients undergoing intracranial pressure and brain tissue oxygen monitoring. Custom algorithms were used to identify periods of elevated ICP, compute burst suppression ratios (BSRs), and detect faulty or disconnected EEG electrodes. Hospital staff were notified of clinically relevant events using our secure API to route alerts through a password-protected smartphone application.

RESULTS

Sustained increases in ICP and concordant decreases in PbtO2 were reliably detected using user-defined thresholds and alert throttling. BSR trends computed by the platform correlated highly with manual neurologist markings (r2 0.633-0.781; p < 0.0001). The platform identified EEG electrodes with poor signal quality with 95% positive predictive value, and reduced latency of technician response by 93%.

CONCLUSION

This study validates a flexible real-time platform for monitoring and interpreting ICU data and notifying caretakers of actionable results, with potential to reduce the manual burden of continuous monitoring services on care providers.

SIGNIFICANCE

This work represents an important step toward facilitating translational medical data analytics to improve patient care and reduce health care costs.

Neurological Prognostication in Children After Cardiac Arrest

Early after pediatric cardiac arrest, families and care providers struggle with the uncertainty of long-term neurological prognosis. Cardiac arrest characteristics such as location, intra-arrest factors, and postarrest events have been associated with outcome. We paid particular attention to postarrest modalities that have been shown to predict neurological outcome. These modalities include neurological examination, somatosensory evoked potentials, electroencephalography, and neuroimaging. There is no one modality that accurately predicts neurological prognosis. Thus, a multimodal approach should be undertaken by both neurologists and intensivists to present a clear and consistent message to families. Methods used for the prediction of long-term neurological prognosis need to be specific enough to identify indivuals with a poor outcome. We review the evidence evaluating children with coma, each with various etiologies of cardiac arrest, outcome measures, and timing of follow-up.

Continuous Electroencephalography Monitoring in Critically Ill Infants and Children

Continuous video electroencephalography (CEEG) monitoring of critically ill infants and children has expanded rapidly in recent years. Indications for CEEG include evaluation of patients with altered mental status, characterization of paroxysmal events, and detection of electrographic seizures, including monitoring of patients with limited neurological examination or conditions that put them at high risk for electrographic seizures (e.g., cardiac arrest or extracorporeal membrane oxygenation cannulation). Depending on the inclusion criteria and clinical characteristics of the population studied, the percentage of pediatric patients with electrographic seizures varies from 7% to 46% and with electrographic status epilepticus from 1% to 23%. There is also evidence that epileptiform and background CEEG patterns may provide important information about prognosis in certain clinical populations. Quantitative EEG techniques are emerging as a tool to enhance the value of CEEG to provide real-time bedside data for management and prognosis. Continued research is needed to understand the clinical value of seizure detection and identification of other CEEG patterns on the outcomes of critically ill infants and children.