Modulation of brain activity in brain-injured patients with a disorder of consciousness in intensive care with repeated 10-Hz transcranial alternating current stimulation (tACS): a randomised controlled trial protocol.

INTRODUCTION: Therapeutic interventions for disorders of consciousness lack consistency; evidence supports non-invasive brain stimulation, but few studies assess neuromodulation in acute-to-subacute brain-injured patients. This study aims to validate the feasibility and assess the effect of a multi-session transcranial alternating current stimulation (tACS) intervention in subacute brain-injured patients on recovery of consciousness, related brain oscillations and brain network dynamics. METHODS AND ANALYSES: The study is comprised of two phases: a validation phase (n=12) and a randomised controlled trial (n=138). Both phases will be conducted in medically stable brain-injured adult patients (traumatic brain injury and hypoxic-ischaemic encephalopathy), with a Glasgow Coma Scale score ≤12 after continuous sedation withdrawal. Recruitment will occur at the intensive care unit of a Level 1 Trauma Centre in Montreal, Quebec, Canada. The intervention includes a 20 min 10 Hz tACS at 1 mA intensity or a sham session over parieto-occipital cortical sites, repeated over five consecutive days. The current's frequency targets alpha brain oscillations (8-13 Hz), known to be associated with consciousness. Resting-state electroencephalogram (EEG) will be recorded four times daily for five consecutive days: pre and post-intervention, at 60 and 120 min post-tACS. Two additional recordings will be included: 24 hours and 1-week post-protocol. Multimodal measures (blood samples, pupillometry, behavioural consciousness assessments (Coma Recovery Scale-revised), actigraphy measures) will be acquired from baseline up to 1 week after the stimulation. EEG signal analysis will focus on the alpha bandwidth (8-13 Hz) using spectral and functional network analyses. Phone assessments at 3, 6 and 12 months post-tACS, will measure long-term functional recovery, quality of life and caregivers' burden. ETHICS AND DISSEMINATION: Ethical approval for this study has been granted by the Research Ethics Board of the CIUSSS du Nord-de-l'Île-de-Montréal (Project ID 2021-2279). The findings of this two-phase study will be submitted for publication in a peer-reviewed academic journal and submitted for presentation at conferences. The trial's results will be published on a public trial registry database (ClinicalTrials.gov). TRIAL REGISTRATION NUMBER: NCT05833568.

Criticality of resting-state EEG predicts perturbational complexity and level of consciousness during anesthesia.

Consciousness has been proposed to be supported by electrophysiological patterns poised at criticality, a dynamical regime which exhibits adaptive computational properties, maximally complex patterns and divergent sensitivity to perturbation. Here, we investigated dynamical properties of the resting-state electroencephalogram of healthy subjects undergoing general anesthesia with propofol, xenon or ketamine. We then studied the relation of these dynamic properties with the perturbational complexity index (PCI), which has shown remarkably high sensitivity in detecting consciousness independent of behavior. All participants were unresponsive under anesthesia, while consciousness was retained only during ketamine anesthesia (in the form of vivid dreams)., enabling an experimental dissociation between unresponsiveness and unconsciousness. We estimated (i) avalanche criticality, (ii) chaoticity, and (iii) criticality-related measures, and found that states of unconsciousness were characterized by a distancing from both the edge of activity propagation and the edge of chaos. We were then able to predict individual subjects' PCI (i.e., PCImax) with a mean absolute error below 7%. Our results establish a firm link between the PCI and criticality and provide further evidence for the role of criticality in the emergence of consciousness.

Class imbalance should not throw you off balance: Choosing the right classifiers and performance metrics for brain decoding with imbalanced data.

Machine learning (ML) is increasingly used in cognitive, computational and clinical neuroscience. The reliable and efficient application of ML requires a sound understanding of its subtleties and limitations. Training ML models on datasets with imbalanced classes is a particularly common problem, and it can have severe consequences if not adequately addressed. With the neuroscience ML user in mind, this paper provides a didactic assessment of the class imbalance problem and illustrates its impact through systematic manipulation of data imbalance ratios in (i) simulated data and (ii) brain data recorded with electroencephalography (EEG), magnetoencephalography (MEG) and functional magnetic resonance imaging (fMRI). Our results illustrate how the widely-used Accuracy (Acc) metric, which measures the overall proportion of successful predictions, yields misleadingly high performances, as class imbalance increases. Because Acc weights the per-class ratios of correct predictions proportionally to class size, it largely disregards the performance on the minority class. A binary classification model that learns to systematically vote for the majority class will yield an artificially high decoding accuracy that directly reflects the imbalance between the two classes, rather than any genuine generalizable ability to discriminate between them. We show that other evaluation metrics such as the Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC), and the less common Balanced Accuracy (BAcc) metric - defined as the arithmetic mean between sensitivity and specificity, provide more reliable performance evaluations for imbalanced data. Our findings also highlight the robustness of Random Forest (RF), and the benefits of using stratified cross-validation and hyperprameter optimization to tackle data imbalance. Critically, for neuroscience ML applications that seek to minimize overall classification error, we recommend the routine use of BAcc, which in the specific case of balanced data is equivalent to using standard Acc, and readily extends to multi-class settings. Importantly, we present a list of recommendations for dealing with imbalanced data, as well as open-source code to allow the neuroscience community to replicate and extend our observations and explore alternative approaches to coping with imbalanced data.

Aperiodic brain activity and response to anesthesia vary in disorders of consciousness.

In the human electroencephalogram (EEG), oscillatory power co-exist with non-oscillatory, aperiodic activity. Although EEG analysis has traditionally focused exclusively on oscillatory power, recent investigations have shown that the aperiodic EEG component can distinguish conscious wakefulness from sleep and anesthetic-induced unconsciousness. This study investigates the aperiodic EEG component of individuals in a disorder of consciousness (DOC); how it changes in response to exposure to anesthesia; and how it relates to the brain's information richness and criticality. High-density EEG was recorded from 43 individuals in a DOC, with 16 of these individuals undergoing a protocol of propofol anesthesia. The aperiodic component was defined by the spectral slope of the power spectral density. Our results demonstrate that the EEG aperiodic component is more informative about the participants' level of consciousness than the oscillatory component, especially for patients that suffered from a stroke. Importantly, the pharmacologically induced change in the spectral slope from 30 to 45 Hz positively correlated with individual's pre-anesthetic level of consciousness. The pharmacologically induced loss of information-richness and criticality was associated with individual's pre-anesthetic aperiodic component. During exposure to anesthesia, the aperiodic component distinguished individuals with DOC, according to their 3-month recovery status. The aperiodic EEG component has been historically neglected; this research highlights the necessity of considering this measure for the assessment of individuals in DOC and future research that seeks to understand the neurophysiological underpinnings of consciousness.

Paradoxical markers of conscious levels: Effects of propofol on patients in disorders of consciousness.

Human consciousness is widely understood to be underpinned by rich and diverse functional networks, whose breakdown results in unconsciousness. Candidate neural correlates of anesthetic-induced unconsciousness include: (1) disrupted frontoparietal functional connectivity; (2) disrupted brain network hubs; and (3) reduced spatiotemporal complexity. However, emerging counterexamples have revealed that these markers may appear outside of the state they are associated with, challenging both their inclusion as markers of conscious level, and the theories of consciousness that rely on their evidence. In this study, we present a case series of three individuals in disorders of consciousness (DOC) who exhibit paradoxical brain responses to exposure to anesthesia. High-density electroencephalographic data were recorded from three patients with unresponsive wakefulness syndrome (UWS) while they underwent a protocol of propofol anesthesia with a targeted effect site concentration of 2 µg/ml. Network hubs and directionality of functional connectivity in the alpha frequency band (8-13 Hz), were estimated using the weighted phase lag index (wPLI) and directed phase lag index (dPLI). The spatiotemporal signal complexity was estimated using three types of Lempel-Ziv complexity (LZC). Our results illustrate that exposure to propofol anesthesia can paradoxically result in: (1) increased frontoparietal feedback-dominant connectivity; (2) posterior network hubs; and (3) increased spatiotemporal complexity. The case examples presented in this paper challenge the role of functional connectivity and spatiotemporal complexity in theories of consciousness and for the clinical evaluation of levels of human consciousness.

Brain Responses to Propofol in Advance of Recovery from Coma and Disorders of Consciousness: A Preliminary Study.

Rationale: Predicting recovery of consciousness in unresponsive, brain-injured individuals has crucial implications for clinical decision-making. Propofol induces distinctive brain network reconfiguration in the healthy brain as it loses consciousness. In patients with disorders of consciousness, the brain network's reconfiguration to propofol may reveal the patient's underlying capacity for consciousness. Objectives: To design and test a new metric for the prognostication of consciousness recovery in disorders of consciousness. Methods: Using a within-subject design, we conducted an anesthetic protocol with concomitant high-density EEG in 12 patients with a disorder of consciousness after a brain injury. We quantified the reconfiguration of EEG network hubs and directed functional connectivity before, during, and after propofol exposure and obtained an index of propofol-induced network reconfiguration: the adaptive reconfiguration index. We compared the index of patients who recovered consciousness 3 months after EEG (n = 3) to that of patients who did not recover or remained in a chronic disorder of consciousness (n = 7) and conducted a logistic regression to assess prognostic accuracy. Measurements and Main Results: The adaptive reconfiguration index was significantly higher in patients who later recovered full consciousness (U value = 21, P = 0.008) and able to discriminate with 100% accuracy whether the patient recovered consciousness. Conclusions: The adaptive reconfiguration index of patients who recovered from a disorder of consciousness at 3-month follow-up was linearly separable from that of patients who did not recover or remained in a chronic disorder of consciousness on the single-subject level. EEG and propofol can be administered at the bedside with few contraindications, affording the adaptive reconfiguration index tremendous translational potential as a prognostic measure of consciousness recovery in acute clinical settings.

Differential classification of states of consciousness using envelope- and phase-based functional connectivity.

The development of sophisticated computational tools to quantify changes in the brain's oscillatory dynamics across states of consciousness have included both envelope- and phase-based measures of functional connectivity (FC), but there are very few direct comparisons of these techniques using the same dataset. The goal of this study was to compare an envelope-based (i.e. Amplitude Envelope Correlation, AEC) and a phase-based (i.e. weighted Phase Lag Index, wPLI) measure of FC in their classification of states of consciousness. Nine healthy participants underwent a three-hour experimental anesthetic protocol with propofol induction and isoflurane maintenance, in which five minutes of 128-channel electroencephalography were recorded before, during, and after anesthetic-induced unconsciousness, at the following time points: Baseline; light sedation with propofol (Light Sedation); deep unconsciousness following three hours of surgical levels of anesthesia with isoflurane (Unconscious); five minutes prior to the recovery of consciousness (Pre-ROC); and three hours following the recovery of consciousness (Recovery). Support vector machine classification was applied to the source-localized EEG in the alpha (8-13 Hz) frequency band in order to investigate the ability of AEC and wPLI (separately and together) to discriminate i) the four states from Baseline; ii) Unconscious ("deep" unconsciousness) vs. Pre-ROC ("light" unconsciousness); and iii) responsiveness (Baseline, Light Sedation, Recovery) vs. unresponsiveness (Unconscious, Pre-ROC). AEC and wPLI yielded different patterns of global connectivity across states of consciousness, with AEC showing the strongest network connectivity during the Unconscious epoch, and wPLI showing the strongest connectivity during full consciousness (i.e., Baseline and Recovery). Both measures also demonstrated differential predictive contributions across participants and used different brain regions for classification. AEC showed higher classification accuracy overall, particularly for distinguishing anesthetic-induced unconsciousness from Baseline (83.7 ± 0.8%). AEC also showed stronger classification accuracy than wPLI when distinguishing Unconscious from Pre-ROC (i.e., "deep" from "light" unconsciousness) (AEC: 66.3 ± 1.2%; wPLI: 56.2 ± 1.3%), and when distinguishing between responsiveness and unresponsiveness (AEC: 76.0 ± 1.3%; wPLI: 63.6 ± 1.8%). Classification accuracy was not improved compared to AEC when both AEC and wPLI were combined. This analysis of source-localized EEG data demonstrates that envelope- and phase-based FC provide different information about states of consciousness but that, on a group level, AEC is better able to detect relative alterations in brain FC across levels of anesthetic-induced unconsciousness compared to wPLI.

Protocol for the Prognostication of Consciousness Recovery Following a Brain Injury.

Individuals who have suffered a severe brain injury typically require extensive hospitalization in intensive care units (ICUs), where critical treatment decisions are made to maximize their likelihood of recovering consciousness and cognitive function. These treatment decisions can be difficult when the neurological assessment of the patient is limited by unreliable behavioral responses. Reliable objective and quantifiable markers are lacking and there is both (1) a poor understanding of the mechanisms underlying the brain's ability to reconstitute consciousness and cognition after an injury and (2) the absence of a reliable and clinically feasible method of tracking cognitive recovery in ICU survivors. Our goal is to develop and validate a clinically relevant EEG paradigm that can inform the prognosis of unresponsive, brain-injured patients in the ICU. This protocol describes a study to develop a point-of-care system intended to accurately predict outcomes of unresponsive, brain-injured patients in the ICU. We will recruit 200 continuously-sedated brain-injured patients across five ICUs. Between 24 h and 7 days post-ICU admission, high-density EEG will be recorded from behaviorally unresponsive patients before, during and after a brief cessation of pharmacological sedation. Once patients have reached the waking stage, they will be asked to complete an abridged Cambridge Brain Sciences battery, a web-based series of neurocognitive tests. The test series will be repeated every day during acute admission (ICU, ward), or as often as possible given the constraints of ICU and ward care. Following discharge, patients will continue to complete the same test series on weekly, and then monthly basis, for up to 12 months following injury. Functional outcomes will also be assessed up to 12 months post-injury. We anticipate our findings will lead to an increased ability to identify patients, as soon as possible after their brain injury, who are most likely to survive, and to make accurate predictions about their long-term cognitive and functional outcome. In addition to providing critically needed support for clinical decision-making, this study has the potential to transform our understanding of key functional EEG networks associated with consciousness and cognition.