Predictable and unpredictable deviance detection in the human hippocampus and amygdala.

Our brains extract structure from the environment and form predictions given past experience. Predictive circuits have been identified in wide-spread cortical regions. However, the contribution of medial temporal structures in predictions remains under-explored. The hippocampus underlies sequence detection and is sensitive to novel stimuli, sufficient to gain access to memory, while the amygdala to novelty. Yet, their electrophysiological profiles in detecting predictable and unpredictable deviant auditory events remain unknown. Here, we hypothesized that the hippocampus would be sensitive to predictability, while the amygdala to unexpected deviance. We presented epileptic patients undergoing presurgical monitoring with standard and deviant sounds, in predictable or unpredictable contexts. Onsets of auditory responses and unpredictable deviance effects were detected earlier in the temporal cortex compared with the amygdala and hippocampus. Deviance effects in 1-20 Hz local field potentials were detected in the lateral temporal cortex, irrespective of predictability. The amygdala showed stronger deviance in the unpredictable context. Low-frequency deviance responses in the hippocampus (1-8 Hz) were observed in the predictable but not in the unpredictable context. Our results reveal a distributed network underlying the generation of auditory predictions and suggest that the neural basis of sensory predictions and prediction error signals needs to be extended.

Intrinsic Neural Timescales in the Temporal Lobe Support an Auditory Processing Hierarchy.

During rest, intrinsic neural dynamics manifest at multiple timescales, which progressively increase along visual and somatosensory hierarchies. Theoretically, intrinsic timescales are thought to facilitate processing of external stimuli at multiple stages. However, direct links between timescales at rest and sensory processing, as well as translation to the auditory system are lacking. Here, we measured intracranial EEG in 11 human patients with epilepsy (4 women), while listening to pure tones. We show that, in the auditory network, intrinsic neural timescales progressively increase, while the spectral exponent flattens, from temporal to entorhinal cortex, hippocampus, and amygdala. Within the neocortex, intrinsic timescales exhibit spatial gradients that follow the temporal lobe anatomy. Crucially, intrinsic timescales at baseline can explain the latency of auditory responses: as intrinsic timescales increase, so do the single-electrode response onset and peak latencies. Our results suggest that the human auditory network exhibits a repertoire of intrinsic neural dynamics, which manifest in cortical gradients with millimeter resolution and may provide a variety of temporal windows to support auditory processing.SIGNIFICANCE STATEMENT Endogenous neural dynamics are often characterized by their intrinsic timescales. These are thought to facilitate processing of external stimuli. However, a direct link between intrinsic timing at rest and sensory processing is missing. Here, with intracranial EEG, we show that intrinsic timescales progressively increase from temporal to entorhinal cortex, hippocampus, and amygdala. Intrinsic timescales at baseline can explain the variability in the timing of intracranial EEG responses to sounds: cortical electrodes with fast timescales also show fast- and short-lasting responses to auditory stimuli, which progressively increase in the hippocampus and amygdala. Our results suggest that a hierarchy of neural dynamics in the temporal lobe manifests across cortical and limbic structures and can explain the temporal richness of auditory responses.

Neural complexity and the spectral slope characterise auditory processing in wakefulness and sleep.

Auditory processing and the complexity of neural activity can both indicate residual consciousness levels and differentiate states of arousal. However, how measures of neural signal complexity manifest in neural activity following environmental stimulation and, more generally, how the electrophysiological characteristics of auditory responses change in states of reduced consciousness remain under-explored. Here, we tested the hypothesis that measures of neural complexity and the spectral slope would discriminate stages of sleep and wakefulness not only in baseline electroencephalography (EEG) activity but also in EEG signals following auditory stimulation. High-density EEG was recorded in 21 participants to determine the spatial relationship between these measures and between EEG recorded pre- and post-auditory stimulation. Results showed that the complexity and the spectral slope in the 2-20 Hz range discriminated between sleep stages and had a high correlation in sleep. In wakefulness, complexity was strongly correlated to the 20-40 Hz spectral slope. Auditory stimulation resulted in reduced complexity in sleep compared to the pre-stimulation EEG activity and modulated the spectral slope in wakefulness. These findings confirm our hypothesis that electrophysiological markers of arousal are sensitive to sleep/wake states in EEG activity during baseline and following auditory stimulation. Our results have direct applications to studies using auditory stimulation to probe neural functions in states of reduced consciousness.

Disentangling the complex landscape of sleep-wake disorders with data-driven phenotyping: A study of the Bernese center.

BACKGROUND AND PURPOSE: The diagnosis of sleep-wake disorders (SWDs) is challenging because of the existence of only few accurate biomarkers and the frequent coexistence of multiple SWDs and/or other comorbidities. The aim of this study was to assess in a large cohort of well-characterized SWD patients the potential of a data-driven approach for the identification of SWDs. METHODS: We included 6958 patients from the Bernese Sleep Registry and 300 variables/biomarkers including questionnaires, results of polysomnography/vigilance tests, and final clinical diagnoses. A pipeline, based on machine learning, was created to extract and cluster the clinical data. Our analysis was performed on three cohorts: patients with central disorders of hypersomnolence (CDHs), a full cohort of patients with SWDs, and a clean cohort without coexisting SWDs. RESULTS: A first analysis focused on the cohort of patients with CDHs and revealed four patient clusters: two clusters for narcolepsy type 1 (NT1) but not for narcolepsy type 2 or idiopathic hypersomnia. In the full cohort of SWDs, nine clusters were found: four contained patients with obstructive and central sleep apnea syndrome, one with NT1, and four with intermixed SWDs. In the cohort of patients without coexisting SWDs, an additional cluster of patients with chronic insomnia disorder was identified. CONCLUSIONS: This study confirms the existence of clear clusters of NT1 in CDHs, but mainly intermixed groups in the full spectrum of SWDs, with the exception of sleep apnea syndromes and NT1. New biomarkers are needed for better phenotyping and diagnosis of SWDs.

Auditory stimulation and deep learning predict awakening from coma after cardiac arrest.

Assessing the integrity of neural functions in coma after cardiac arrest remains an open challenge. Prognostication of coma outcome relies mainly on visual expert scoring of physiological signals, which is prone to subjectivity and leaves a considerable number of patients in a 'grey zone', with uncertain prognosis. Quantitative analysis of EEG responses to auditory stimuli can provide a window into neural functions in coma and information about patients' chances of awakening. However, responses to standardized auditory stimulation are far from being used in a clinical routine due to heterogeneous and cumbersome protocols. Here, we hypothesize that convolutional neural networks can assist in extracting interpretable patterns of EEG responses to auditory stimuli during the first day of coma that are predictive of patients' chances of awakening and survival at 3 months. We used convolutional neural networks (CNNs) to model single-trial EEG responses to auditory stimuli in the first day of coma, under standardized sedation and targeted temperature management, in a multicentre and multiprotocol patient cohort and predict outcome at 3 months. The use of CNNs resulted in a positive predictive power for predicting awakening of 0.83 ± 0.04 and 0.81 ± 0.06 and an area under the curve in predicting outcome of 0.69 ± 0.05 and 0.70 ± 0.05, for patients undergoing therapeutic hypothermia and normothermia, respectively. These results also persisted in a subset of patients that were in a clinical 'grey zone'. The network's confidence in predicting outcome was based on interpretable features: it strongly correlated to the neural synchrony and complexity of EEG responses and was modulated by independent clinical evaluations, such as the EEG reactivity, background burst-suppression or motor responses. Our results highlight the strong potential of interpretable deep learning algorithms in combination with auditory stimulation to improve prognostication of coma outcome.

Asymmetric representation of aversive prediction errors in Pavlovian threat conditioning.

Survival in biological environments requires learning associations between predictive sensory cues and threatening outcomes. Such aversive learning may be implemented through reinforcement learning algorithms that are driven by the signed difference between expected and encountered outcomes, termed prediction errors (PEs). While PE-based learning is well established for reward learning, the role of putative PE signals in aversive learning is less clear. Here, we used functional magnetic resonance imaging in humans (21 healthy men and women) to investigate the neural representation of PEs during maintenance of learned aversive associations. Four visual cues, each with a different probability (0, 33, 66, 100%) of being followed by an aversive outcome (electric shock), were repeatedly presented to participants. We found that neural activity at omission (US-) but not occurrence of the aversive outcome (US+) encoded PEs in the medial prefrontal cortex. More expected omission of aversive outcome was associated with lower neural activity. No neural signals fulfilled axiomatic criteria, which specify necessary and sufficient components of PE signals, for signed PE representation in a whole-brain search or in a-priori regions of interest. Our results might suggest that, different from reward learning, aversive learning does not involve signed PE signals that are represented within the same brain region for all conditions.

Complementary roles of neural synchrony and complexity for indexing consciousness and chances of surviving in acute coma.

An open challenge in consciousness research is understanding how neural functions are altered by pathological loss of consciousness. To maintain consciousness, the brain needs synchronized communication of information across brain regions, and sufficient complexity in neural activity. Coordination of brain activity, typically indexed through measures of neural synchrony, has been shown to decrease when consciousness is lost and to reflect the clinical state of patients with disorders of consciousness. Moreover, when consciousness is lost, neural activity loses complexity, while the levels of neural noise, indexed by the slope of the electroencephalography (EEG) spectral exponent decrease. Although these properties have been well investigated in resting state activity, it remains unknown whether the sensory processing network, which has been shown to be preserved in coma, suffers from a loss of synchronization or information content. Here, we focused on acute coma and hypothesized that neural synchrony in response to auditory stimuli would reflect coma severity, while complexity, or neural noise, would reflect the presence or loss of consciousness. Results showed that neural synchrony of EEG signals was stronger for survivors than non-survivors and predictive of patients' outcome, but indistinguishable between survivors and healthy controls. Measures of neural complexity and neural noise were not informative of patients' outcome and had high or low values for patients compared to controls. Our results suggest different roles for neural synchrony and complexity in acute coma. Synchrony represents a precondition for consciousness, while complexity needs an equilibrium between high or low values to support conscious cognition.

Embracing diversity and inclusivity in an academic setting: Insights from the Organization for Human Brain Mapping.

Scientific research aims to bring forward innovative ideas and constantly challenges existing knowledge structures and stereotypes. However, women, ethnic and cultural minorities, as well as individuals with disabilities, are systematically discriminated against or even excluded from promotions, publications, and general visibility. A more diverse workforce is more productive, and thus discrimination has a negative impact on science and the wider society, as well as on the education, careers, and well-being of individuals who are discriminated against. Moreover, the lack of diversity at scientific gatherings can lead to micro-aggressions or harassment, making such meetings unpleasant, or even unsafe environments for early career and underrepresented scientists. At the Organization for Human Brain Mapping (OHBM), we recognized the need for promoting underrepresented scientists and creating diverse role models in the field of neuroimaging. To foster this, the OHBM has created a Diversity and Inclusivity Committee (DIC). In this article, we review the composition and activities of the DIC that have promoted diversity within OHBM, in order to inspire other organizations to implement similar initiatives. Activities of the committee over the past four years have included (a) creating a code of conduct, (b) providing diversity and inclusivity education for OHBM members, (c) organizing interviews and symposia on diversity issues, and (d) organizing family-friendly activities and providing childcare grants during the OHBM annual meetings. We strongly believe that these activities have brought positive change within the wider OHBM community, improving inclusivity and fostering diversity while promoting rigorous, ground-breaking science. These positive changes could not have been so rapidly implemented without the enthusiastic support from the leadership, including OHBM Council and Program Committee, and the OHBM Special Interest Groups (SIGs), namely the Open Science, Student and Postdoc, and Brain-Art SIGs. Nevertheless, there remains ample room for improvement, in all areas, and even more so in the area of targeted attempts to increase inclusivity for women, individuals with disabilities, members of the LGBTQ+ community, racial/ethnic minorities, and individuals of lower socioeconomic status or from low and middle-income countries. Here, we present an overview of the DIC's composition, its activities, future directions and challenges. Our goal is to share our experiences with a wider audience to provide information to other organizations and institutions wishing to implement similar comprehensive diversity initiatives. We propose that scientific organizations can push the boundaries of scientific progress only by moving beyond existing power structures and by integrating principles of equity and inclusivity in their core values.

The Swiss Primary Hypersomnolence and Narcolepsy Cohort study (SPHYNCS): Study protocol for a prospective, multicentre cohort observational study.

Narcolepsy type 1 (NT1) is a disorder with well-established markers and a suspected autoimmune aetiology. Conversely, the narcoleptic borderland (NBL) disorders, including narcolepsy type 2, idiopathic hypersomnia, insufficient sleep syndrome and hypersomnia associated with a psychiatric disorder, lack well-defined markers and remain controversial in terms of aetiology, diagnosis and management. The Swiss Primary Hypersomnolence and Narcolepsy Cohort Study (SPHYNCS) is a comprehensive multicentre cohort study, which will investigate the clinical picture, pathophysiology and long-term course of NT1 and the NBL. The primary aim is to validate new and reappraise well-known markers for the characterization of the NBL, facilitating the diagnostic process. Seven Swiss sleep centres, belonging to the Swiss Narcolepsy Network (SNaNe), joined the study and will prospectively enrol over 500 patients with recent onset of excessive daytime sleepiness (EDS), hypersomnia or a suspected central disorder of hypersomnolence (CDH) during a 3-year recruitment phase. Healthy controls and patients with EDS due to severe sleep-disordered breathing, improving after therapy, will represent two control groups of over 50 patients each. Clinical and electrophysiological (polysomnography, multiple sleep latency test, maintenance of wakefulness test) information, and information on psychomotor vigilance and a sustained attention to response task, actigraphy and wearable devices (long-term monitoring), and responses to questionnaires will be collected at baseline and after 6, 12, 24 and 36 months. Potential disease markers will be searched for in blood, cerebrospinal fluid and stool. Analyses will include quantitative hypocretin measurements, proteomics/peptidomics, and immunological, genetic and microbiota studies. SPHYNCS will increase our understanding of CDH and the relationship between NT1 and the NBL. The identification of new disease markers is expected to lead to better and earlier diagnosis, better prognosis and personalized management of CDH.

The relation between neuronal firing, local field potentials and hemodynamic activity in the human amygdala in response to aversive dynamic visual stimuli.

The amygdala is a central part of networks of brain regions underlying perception and cognition, in particular related to processing of emotionally salient stimuli. Invasive electrophysiological and hemodynamic measurements are commonly used to evaluate functions of the human amygdala, but a comprehensive understanding of their relation is still lacking. Here, we aimed at investigating the link between fast and slow frequency amygdalar oscillations, neuronal firing and hemodynamic responses. To this aim, we recorded intracranial electroencephalography (iEEG), hemodynamic responses and single neuron activity from the amygdala of patients with epilepsy. Patients were presented with dynamic visual sequences of fearful faces (aversive condition), interleaved with sequences of neutral landscapes (neutral condition). Comparing responses to aversive versus neutral stimuli across participants, we observed enhanced high gamma power (HGP, >60 â€‹Hz) during the first 2 â€‹s of aversive sequence viewing, and reduced delta power (1-4 â€‹Hz) lasting up to 18 â€‹s. In 5 participants with implanted microwires, neuronal firing rates were enhanced following aversive stimuli, and exhibited positive correlation with HGP and hemodynamic responses. Our results show that high gamma power, neuronal firing and BOLD responses from the human amygdala are co-modulated. Our findings provide, for the first time, a comprehensive investigation of amygdalar responses to aversive stimuli, ranging from single-neuron spikes to local field potentials and hemodynamic responses.

High-precision magnetoencephalography for reconstructing amygdalar and hippocampal oscillations during prediction of safety and threat.

Learning to associate neutral with aversive events in rodents is thought to depend on hippocampal and amygdala oscillations. In humans, oscillations underlying aversive learning are not well characterised, largely due to the technical difficulty of recording from these two structures. Here, we used high-precision magnetoencephalography (MEG) during human discriminant delay threat conditioning. We constructed generative anatomical models relating neural activity with recorded magnetic fields at the single-participant level, including the neocortex with or without the possibility of sources originating in the hippocampal and amygdalar structures. Models including neural activity in amygdala and hippocampus explained MEG data during threat conditioning better than exclusively neocortical models. We found that in both amygdala and hippocampus, theta oscillations during anticipation of an aversive event had lower power compared to safety, both during retrieval and extinction of aversive memories. At the same time, theta synchronisation between hippocampus and amygdala increased over repeated retrieval of aversive predictions, but not during safety. Our results suggest that high-precision MEG is sensitive to neural activity of the human amygdala and hippocampus during threat conditioning and shed light on the oscillation-mediated mechanisms underpinning retrieval and extinction of fear memories in humans.

Human Pavlovian fear conditioning conforms to probabilistic learning.

Learning to predict threat from environmental cues is a fundamental skill in changing environments. This aversive learning process is exemplified by Pavlovian threat conditioning. Despite a plethora of studies on the neural mechanisms supporting the formation of associations between neutral and aversive events, our computational understanding of this process is fragmented. Importantly, different computational models give rise to different and partly opposing predictions for the trial-by-trial dynamics of learning, for example expressed in the activity of the autonomic nervous system (ANS). Here, we investigate human ANS responses to conditioned stimuli during Pavlovian fear conditioning. To obtain precise, trial-by-trial, single-subject estimates of ANS responses, we build on a statistical framework for psychophysiological modelling. We then consider previously proposed non-probabilistic models, a simple probabilistic model, and non-learning models, as well as different observation functions to link learning models with ANS activity. Across three experiments, and both for skin conductance (SCR) and pupil size responses (PSR), a probabilistic learning model best explains ANS responses. Notably, SCR and PSR reflect different quantities of the same model: SCR track a mixture of expected outcome and uncertainty, while PSR track expected outcome alone. In summary, by combining psychophysiological modelling with computational learning theory, we provide systematic evidence that the formation and maintenance of Pavlovian threat predictions in humans may rely on probabilistic inference and includes estimation of uncertainty. This could inform theories of neural implementation of aversive learning.

Prediction of awakening from hypothermic postanoxic coma based on auditory discrimination.

OBJECTIVE: Most of the available clinical tests for prognosis of postanoxic coma are informative of poor outcome. Previous work has shown that an improvement in auditory discrimination over the first days of coma is predictive of awakening. Here, we aimed at evaluating this test on a large cohort of patients undergoing therapeutic hypothermia and at investigating its added value on existing clinical measures. METHODS: We recorded electroencephalographic responses to auditory stimuli in 94 comatose patients, under hypothermia and after rewarming to normal temperature. Auditory discrimination was semiautomatically quantified by decoding electroencephalographic responses to frequently repeated versus rare sounds. Outcome prediction was based on the change of decoding performance from hypothermia to normothermia. RESULTS: An increase in auditory discrimination from hypothermia to normothermia was observed for 33 of 94 patients. Among them, 27 awoke from coma, resulting in a positive predictive value of awakening of 82% (95% confidence interval = 0.65-0.93). Most nonsurvivors showing an improvement in auditory discrimination had incident status epilepticus. By excluding them, 27 of 29 patients with improvement in auditory discrimination survived, resulting in a considerable improvement of the predictive value for awakening (93%, with 95% confidence interval = 0.77-0.99). Importantly, this test predicted the awakening of 13 of 51 patients for whom the outcome was uncertain based on current tests. INTERPRETATION: The progression of auditory discrimination from hypothermia to normothermia has a high predictive value for awakening. This quantitative measure provides an added value to existing clinical tests and encourages the maintenance of life support. Ann Neurol 2016;79:748-757.

Evidence of trace conditioning in comatose patients revealed by the reactivation of EEG responses to alerting sounds.

Trace conditioning refers to a learning process occurring after repeated presentation of a neutral conditioned stimulus (CS+) and a salient unconditioned stimulus (UCS) separated by a temporal gap. Recent studies have reported that trace conditioning can occur in humans in reduced levels of consciousness by showing a transfer of the unconditioned autonomic response to the CS+ in healthy sleeping individuals and in vegetative state patients. However, no previous studies have investigated the neural underpinning of trace conditioning in the absence of consciousness in humans. In the present study, we recorded the EEG activity of 29 post-anoxic comatose patients while presenting a trace conditioning paradigm using neutral tones as CS+ and alerting sounds as UCS. Most patients received therapeutic hypothermia and all were deeply unconscious according to standardized clinical scales. After repeated presentation of the CS+ and UCS couple, learning was assessed by measuring the EEG activity during the period where the UCS is omitted after CS+ presentation. Specifically we assessed the 'reactivation' of the neural response to UCS omission by applying a decoding algorithm derived from the statistical model of the EEG activity in response to the UCS presentation. The same procedure was used in a group of 12 awake healthy controls. We found a reactivation of the UCS response in absence of stimulation in eight patients (five under therapeutic hypothermia) and four healthy controls. Additionally, the reactivation effect was temporally specific within trials since it manifested primarily at the specific latency of UCS presentation and significantly less before or after this period. Our results show for the first time that trace conditioning may manifest as a reactivation of the EEG activity related to the UCS and even in the absence of consciousness.

Neural detection of complex sound sequences in the absence of consciousness.

The neural response to a violation of sequences of identical sounds is a typical example of the brain's sensitivity to auditory regularities. Previous literature interprets this effect as a pre-attentive and unconscious processing of sensory stimuli. By contrast, a violation to auditory global regularities, i.e. based on repeating groups of sounds, is typically detectable when subjects can consciously perceive them. Here, we challenge the notion that global detection implies consciousness by testing the neural response to global violations in a group of 24 patients with post-anoxic coma (three females, age range 45-87 years), treated with mild therapeutic hypothermia and sedation. By applying a decoding analysis to electroencephalographic responses to standard versus deviant sound sequences, we found above-chance decoding performance in 10 of 24 patients (Wilcoxon signed-rank test, P < 0.001), despite five of them being mildly hypothermic, sedated and unarousable. Furthermore, consistently with previous findings based on the mismatch negativity the progression of this decoding performance was informative of patients' chances of awakening (78% predictive of awakening). Our results show for the first time that detection of global regularities at neural level exists despite a deeply unconscious state.

Experience-based Auditory Predictions Modulate Brain Activity to Silence as do Real Sounds.

Interactions between stimuli's acoustic features and experience-based internal models of the environment enable listeners to compensate for the disruptions in auditory streams that are regularly encountered in noisy environments. However, whether auditory gaps are filled in predictively or restored a posteriori remains unclear. The current lack of positive statistical evidence that internal models can actually shape brain activity as would real sounds precludes accepting predictive accounts of filling-in phenomenon. We investigated the neurophysiological effects of internal models by testing whether single-trial electrophysiological responses to omitted sounds in a rule-based sequence of tones with varying pitch could be decoded from the responses to real sounds and by analyzing the ERPs to the omissions with data-driven electrical neuroimaging methods. The decoding of the brain responses to different expected, but omitted, tones in both passive and active listening conditions was above chance based on the responses to the real sound in active listening conditions. Topographic ERP analyses and electrical source estimations revealed that, in the absence of any stimulation, experience-based internal models elicit an electrophysiological activity different from noise and that the temporal dynamics of this activity depend on attention. We further found that the expected change in pitch direction of omitted tones modulated the activity of left posterior temporal areas 140-200 msec after the onset of omissions. Collectively, our results indicate that, even in the absence of any stimulation, internal models modulate brain activity as do real sounds, indicating that auditory filling in can be accounted for by predictive activity.

Progression of auditory discrimination based on neural decoding predicts awakening from coma.

Auditory evoked potentials are informative of intact cortical functions of comatose patients. The integrity of auditory functions evaluated using mismatch negativity paradigms has been associated with their chances of survival. However, because auditory discrimination is assessed at various delays after coma onset, it is still unclear whether this impairment depends on the time of the recording. We hypothesized that impairment in auditory discrimination capabilities is indicative of coma progression, rather than of the comatose state itself and that rudimentary auditory discrimination remains intact during acute stages of coma. We studied 30 post-anoxic comatose patients resuscitated from cardiac arrest and five healthy, age-matched controls. Using a mismatch negativity paradigm, we performed two electroencephalography recordings with a standard 19-channel clinical montage: the first within 24 h after coma onset and under mild therapeutic hypothermia, and the second after 1 day and under normothermic conditions. We analysed electroencephalography responses based on a multivariate decoding algorithm that automatically quantifies neural discrimination at the single patient level. Results showed high average decoding accuracy in discriminating sounds both for control subjects and comatose patients. Importantly, accurate decoding was largely independent of patients' chance of survival. However, the progression of auditory discrimination between the first and second recordings was informative of a patient's chance of survival. A deterioration of auditory discrimination was observed in all non-survivors (equivalent to 100% positive predictive value for survivors). We show, for the first time, evidence of intact auditory processing even in comatose patients who do not survive and that progression of sound discrimination over time is informative of a patient's chance of survival. Tracking auditory discrimination in comatose patients could provide new insight to the chance of awakening in a quantitative and automatic fashion during early stages of coma.

Bridging AI and Clinical Practice: Integrating Automated Sleep Scoring Algorithm with Uncertainty-Guided Physician Review.

Purpose: This study aims to enhance the clinical use of automated sleep-scoring algorithms by incorporating an uncertainty estimation approach to efficiently assist clinicians in the manual review of predicted hypnograms, a necessity due to the notable inter-scorer variability inherent in polysomnography (PSG) databases. Our efforts target the extent of review required to achieve predefined agreement levels, examining both in-domain (ID) and out-of-domain (OOD) data, and considering subjects' diagnoses. Patients and Methods: A total of 19,578 PSGs from 13 open-access databases were used to train U-Sleep, a state-of-the-art sleep-scoring algorithm. We leveraged a comprehensive clinical database of an additional 8832 PSGs, covering a full spectrum of ages (0-91 years) and sleep-disorders, to refine the U-Sleep, and to evaluate different uncertainty-quantification approaches, including our novel confidence network. The ID data consisted of PSGs scored by over 50 physicians, and the two OOD sets comprised recordings each scored by a unique senior physician. Results: U-Sleep demonstrated robust performance, with Cohen's kappa (K) at 76.2% on ID and 73.8-78.8% on OOD data. The confidence network excelled at identifying uncertain predictions, achieving AUROC scores of 85.7% on ID and 82.5-85.6% on OOD data. Independently of sleep-disorder status, statistical evaluations revealed significant differences in confidence scores between aligning vs discording predictions, and significant correlations of confidence scores with classification performance metrics. To achieve κ ≥ 90% with physician intervention, examining less than 29.0% of uncertain epochs was required, substantially reducing physicians' workload, and facilitating near-perfect agreement. Conclusion: Inter-scorer variability limits the accuracy of the scoring algorithms to ~80%. By integrating an uncertainty estimation with U-Sleep, we enhance the review of predicted hypnograms, to align with the scoring taste of a responsible physician. Validated across ID and OOD data and various sleep-disorders, our approach offers a strategy to boost automated scoring tools' usability in clinical settings.

Feasibility, Safety, and Performance of Full-Head Subscalp EEG Using Minimally Invasive Electrode Implantation.

BACKGROUND AND OBJECTIVES: Current practice in clinical neurophysiology is limited to short recordings with conventional EEG (days) that fail to capture a range of brain (dys)functions at longer timescales (months). The future ability to optimally manage chronic brain disorders, such as epilepsy, hinges upon finding methods to monitor electrical brain activity in daily life. We developed a device for full-head subscalp EEG (Epios) and tested here the feasibility to safely insert the electrode leads beneath the scalp by a minimally invasive technique (primary outcome). As secondary outcome, we verified the noninferiority of subscalp EEG in measuring physiologic brain oscillations and pathologic discharges compared with scalp EEG, the established standard of care. METHODS: Eight participants with pharmacoresistant epilepsy undergoing intracranial EEG received in the same surgery subscalp electrodes tunneled between the scalp and the skull with custom-made tools. Postoperative safety was monitored on an inpatient ward for up to 9 days. Sleep-wake, ictal, and interictal EEG signals from subscalp, scalp, and intracranial electrodes were compared quantitatively using windowed multitaper transforms and spectral coherence. Noninferiority was tested for pairs of neighboring subscalp and scalp electrodes with a Bland-Altman analysis for measurement bias and calculation of the interclass correlation coefficient (ICC). RESULTS: As primary outcome, up to 28 subscalp electrodes could be safely placed over the entire head through 1-cm scalp incisions in a ∼1-hour procedure. Five of 10 observed perioperative adverse events were linked to the investigational procedure, but none were serious, and all resolved. As a secondary outcome, subscalp electrodes advantageously recorded EEG percutaneously without requiring any maintenance and were noninferior to scalp electrodes for measuring (1) variably strong, stage-specific brain oscillations (alpha in wake, delta, sigma, and beta in sleep) and (2) interictal spikes peak-potentials and ictal signals coherent with seizure propagation in different brain regions (ICC >0.8 and absence of bias). DISCUSSION: Recording full-head subscalp EEG for localization and monitoring purposes is feasible up to 9 days in humans using minimally invasive techniques and noninferior to the current standard of care. A longer prospective ambulatory study of the full system will be necessary to establish the safety and utility of this innovative approach. TRIAL REGISTRATION INFORMATION: clinicaltrials.gov/study/NCT04796597.

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.