Prospective evaluation of interrater agreement between EEG technologists and neurophysiologists

Reaching the ceiling? Empirical scaling behaviour for deep EEG pathology classification

Machine learning techniques, particularly deep convolutional neural networks (ConvNets), are increasingly being used to automate clinical EEG analysis, with the potential to reduce the clinical burden and improve patient care. However, further research is required before they can be used in clinical settings, particularly regarding the impact of the number of training samples and model parameters on their testing error. To address this, we present a comprehensive study of the empirical scaling behaviour of ConvNets for EEG pathology classification. We analysed the testing error with increasing the training samples and model size for four different ConvNet architectures. The focus of our experiments is width scaling, and we have increased the number of parameters to up to 1.8 million. Our evaluation was based on two publicly available datasets: the Temple University Hospital (TUH) Abnormal EEG Corpus and the TUH Abnormal Expansion Balanced EEG Corpus, which together contain 10,707 training samples. The results show that the testing error follows a saturating power-law with both model and dataset size. This pattern is consistent across different datasets and ConvNet architectures. Furthermore, empirically observed accuracies saturate at 85%-87%, which may be due to an imperfect inter-rater agreement on the clinical labels. The empirical scaling behaviour of the test performance with dataset and model size has significant implications for deep EEG pathology classification research and practice. Our findings highlight the potential of deep ConvNets for high-performance EEG pathology classification, and the identified scaling relationships provide valuable recommendations for the advancement of automated EEG diagnostics.

Epilepsy monitoring unit practices and safety among NAEC epilepsy centers: A census survey

OBJECTIVE

An epilepsy monitoring unit (EMU) is a specialized unit designed for capturing and characterizing seizures and other paroxysmal events with continuous video electroencephalography (vEEG). Nearly 260 epilepsy centers in the United States are accredited by the National Association of Epilepsy Centers (NAEC) based on adherence to specific clinical standards to improve epilepsy care, safety, and quality. This study examines EMU staffing, safety practices, and reported outcomes.

METHOD

We analyzed NAEC annual report data and results from a supplemental survey specific to EMU practices reported in 2019 from 341 pediatric or adult center directors. Data on staffing, resources, safety practices and complications were collated with epilepsy center characteristics. We summarized using frequency (percentage) for categorical variables and median (inter-quartile range) for continuous variables. We used chi-square or Fisher's exact tests to compare staff responsibilities.

RESULTS

The supplemental survey response rate was 100%. Spell classification (39%) and phase 1 testing (28%) were the most common goals of the 91,069 reported admissions. The goal ratio of EEG technologist to beds of 1:4 was the most common during the day (68%) and off-hours (43%). Compared to residents and fellows, advanced practice providers served more roles in the EMU at level 3 or pediatric-only centers. Status epilepticus (SE) was the most common reported complication (1.6% of admissions), while cardiac arrest occurred in 0.1% of admissions.

SIGNIFICANCE

EMU staffing and safety practices vary across US epilepsy centers. Reported complications in EMUs are rare but could be further reduced, such as with more effective treatment or prevention of SE. These findings have potential implications for improving EMU safety and quality care.

Deep learning for EEG-based prognostication after cardiac arrest: from current research to future clinical applications

Outcome prognostication in comatose patients after cardiac arrest (CA) remains to date a challenge. The major determinant of clinical outcome is the post-hypoxic/ischemic encephalopathy. Electroencephalography (EEG) is routinely used to assess neural functions in comatose patients. Currently, EEG-based outcome prognosis relies on visual evaluation by medical experts, which is time consuming, prone to subjectivity, and oblivious to complex patterns. The field of deep learning has given rise to powerful algorithms for detecting patterns in large amounts of data. Analyzing EEG signals of coma patients with deep neural networks with the goal of assisting in outcome prognosis is therefore a natural application of these algorithms. Here, we provide the first narrative literature review on the use of deep learning for prognostication after CA. Existing studies show overall high performance in predicting outcome, relying either on spontaneous or on auditory evoked EEG signals. Moreover, the literature is concerned with algorithmic interpretability, and has shown that largely, deep neural networks base their decisions on clinically or neurophysiologically meaningful features. We conclude this review by discussing considerations that the fields of artificial intelligence and neurology will need to jointly address in the future, in order for deep learning algorithms to break the publication barrier, and to be integrated in clinical practice.

Repetitive Electroencephalography as Biomarker for the Prediction of Survival in Patients with Post-Hypoxic Encephalopathy

Predicting survival in patients with post-hypoxic encephalopathy (HE) after cardiopulmonary resuscitation is a challenging aspect of modern neurocritical care. Here, continuous electroencephalography (cEEG) has been established as the gold standard for neurophysiological outcome prediction. Unfortunately, cEEG is not comprehensively available, especially in rural regions and developing countries. The objective of this monocentric study was to investigate the predictive properties of repetitive EEGs (rEEGs) with respect to 12-month survival based on data for 199 adult patients with HE, using log-rank and multivariate Cox regression analysis (MCRA). A total number of 59 patients (29.6%) received more than one EEG during the first 14 days of acute neurocritical care. These patients were analyzed for the presence of and changes in specific EEG patterns that have been shown to be associated with favorable or poor outcomes in HE. Based on MCRA, an initially normal amplitude with secondary low-voltage EEG remained as the only significant predictor for an unfavorable outcome, whereas all other relevant parameters identified by univariate analysis remained non-significant in the model. In conclusion, rEEG during early neurocritical care may help to assess the prognosis of HE patients if cEEG is not available.