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.

Alterations of the alpha rhythm in visual snow syndrome: a case-control study.

BACKGROUND: Visual snow syndrome is a disorder characterized by the combination of typical perceptual disturbances. The clinical picture suggests an impairment of visual filtering mechanisms and might involve primary and secondary visual brain areas, as well as higher-order attentional networks. On the level of cortical oscillations, the alpha rhythm is a prominent EEG pattern that is involved in the prioritisation of visual information. It can be regarded as a correlate of inhibitory modulation within the visual network. METHODS: Twenty-one patients with visual snow syndrome were compared to 21 controls matched for age, sex, and migraine. We analysed the resting-state alpha rhythm by identifying the individual alpha peak frequency using a Fast Fourier Transform and then calculating the power spectral density around the individual alpha peak (+/- 1 Hz). We anticipated a reduced power spectral density in the alpha band over the primary visual cortex in participants with visual snow syndrome. RESULTS: There were no significant differences in the power spectral density in the alpha band over the occipital electrodes (O1 and O2), leading to the rejection of our primary hypothesis. However, the power spectral density in the alpha band was significantly reduced over temporal and parietal electrodes. There was also a trend towards increased individual alpha peak frequency in the subgroup of participants without comorbid migraine. CONCLUSIONS: Our main finding was a decreased power spectral density in the alpha band over parietal and temporal brain regions corresponding to areas of the secondary visual cortex. These findings complement previous functional and structural imaging data at a electrophysiological level. They underscore the involvement of higher-order visual brain areas, and potentially reflect a disturbance in inhibitory top-down modulation. The alpha rhythm alterations might represent a novel target for specific neuromodulation. TRIAL REGISTRATION: we preregistered the study before preprocessing and data analysis on the platform osf.org (DOI: https://doi.org/10.17605/OSF.IO/XPQHF , date of registration: November 19th 2022).

A Pilot Investigation Evaluating Relative Changes in Fronto-Occipital Alpha and Beta Spectral Power as Measurement of Anesthesia Hypnotic Depth.

BACKGROUND: Other than clinical observation of a patient's vegetative response to nociception, monitoring the hypnotic component of general anesthesia (GA) and unconsciousness relies on electroencephalography (EEG)-based indices. These indices exclusively based on frontal EEG activity neglect an important observation. One of the main hallmarks of transitions from wakefulness to GA is a shift in alpha oscillations (7.5-12.5 Hz activity) from occipital brain regions toward anterior brain regions ("alpha anteriorization"). Monitoring the degree of this alpha anteriorization may help to guide induction and maintenance of hypnotic depth and prevent intraoperative awareness. However, the occipital region of the brain is completely disregarded and occipital alpha as characteristic of wakefulness and its posterior-to-anterior shift during induction are missed. Here, we propose an application of Narcotrend's reduced power alpha beta (RPAB) index, originally developed to monitor differences in hemispheric perfusion, for determining the ratio of alpha and beta activity in the anterior-posterior axis. METHODS: Perioperative EEG data of 32 patients undergoing GA in the ophthalmic surgery department of Bern University Hospital were retrospectively analyzed. EEG was recorded with the Narcotrend® monitor using a frontal (Fp1-Fp2) and a posterior (T9-Oz) bipolar derivation with reference electrode over A2. The RPAB index was computed between both bipolar signals, defining the fronto-occipital RPAB (FO-RPAB). FO-RPAB was analyzed during wakefulness, GA maintenance, and emergence, as well as before and after the intraoperative administration of a ketamine bolus. FO-RPAB was compared with a classical quantitative EEG measure-the spectral edge frequency 95% (SEF-95). RESULTS: A significant shift of the FO-RPAB was observed during both induction of and emergence from GA ( P < .001). Interestingly, the additional administration of ketamine during GA did not lead to a significant change in FO-RPAB ( P = 0.81). In contrast, a significant increase in the SEF-95 in the frontal channel was observed during the 10-minute period after ketamine administration ( P < .001). CONCLUSIONS: FO-RPAB appears to qualify as a marker of unconsciousness, reflecting physiological fronto-occipital activity differences during GA. In contrast to frontal SEF-95, it is not disturbed by additional administration of ketamine for analgesia.

Thalamic and neocortical differences in the relationship between the time course of delta and sigma power during NREM sleep in humans.

Sleep spindles and slow waves are the hallmarks of non-rapid eye movement (NREM) sleep and are produced by the dynamic interplay between thalamic and cortical regions. Several studies in both human and animal models have focused their attention on the relationship between electroencephalographic (EEG) spindles and slow waves during NREM, using the power in the sigma and delta bands as a surrogate for the production of spindles and slow waves. A typical report is an overall inverse relationship between the time course of sigma and delta power as measured by a single correlation coefficient both within and across NREM episodes. Here we analysed stereotactically implanted intracerebral electrode (Stereo-EEG [SEEG]) recordings during NREM simultaneously acquired from thalamic and from several neocortical sites in six neurosurgical patients. We investigated the relationship between the time course of delta and sigma power and found that, although at the cortical level it shows the expected inverse relationship, these two frequency bands follow a parallel time course at the thalamic level. Both these observations were consistent across patients and across different cortical as well as thalamic regions. These different temporal dynamics at the neocortical and thalamic level are discussed, considering classical as well as more recent interpretations of the neurophysiological determinants of sleep spindles and slow waves. These findings may also help understanding the regulatory mechanisms of these fundamental sleep EEG graphoelements across different brain compartments.

EEG spindles integrity in critical care adults. Analysis of a randomized trial.

OBJECTIVES: Occurrence of EEG spindles has been recently associated with favorable outcome in ICU patients. Available data mostly rely on relatively small patients' samples, particular etiologies, and limited variables ascertainment. We aimed to expand previous findings on a larger dataset, to identify clinical and EEG patterns correlated with spindle occurrence, and explore its prognostic implications. METHODS: Retrospective observational study of prospectively collected data from a randomized trial (CERTA, NCT03129438) assessing the relationship of continuous (cEEG) versus repeated routine EEG (rEEG) with outcome in adults with acute consciousness impairment. Spindles were prospectively assessed visually as 12-16Hz activity on fronto-central midline regions, at any time during EEG interventions. Uni- and multivariable analyses explored correlations between spindles occurrence, clinical and EEG variables, and outcome (modified Rankin Scale, mRS; mortality) at 6 months. RESULTS: Among the analyzed 364 patients, spindles were independently associated with EEG background reactivity (OR 13.2, 95% CI: 3.11-56.26), and cEEG recording (OR 4.35, 95% CI: 2.5 - 7.69). In the cEEG subgroup (n=182), 33.5% had spindles. They had better FOUR scores (p=0.004), fewer seizures or status epilepticus (p=0.02), and lower mRS (p=0.02). Mortality was reduced (p=0.002), and independently inversely associated with spindle occurrence (OR 0.50, CI 95% 0.25-0.99) and increased EEG background continuity (OR 0.16, 95% CI: 0.07 - 0.41). CONCLUSIONS: Besides confirming that spindle activity occurs in up to one third of acutely ill patients and is associated with better outcome, this study shows that cEEG has a higher yield than rEEG in identifying them. Furthermore, it unravels associations with several clinical and EEG features in this clinical setting.

Standardized visual EEG features predict outcome in patients with acute consciousness impairment of various etiologies.

BACKGROUND: Early prognostication in patients with acute consciousness impairment is a challenging but essential task. Current prognostic guidelines vary with the underlying etiology. In particular, electroencephalography (EEG) is the most important paraclinical examination tool in patients with hypoxic ischemic encephalopathy (HIE), whereas it is not routinely used for outcome prediction in patients with traumatic brain injury (TBI). METHOD: Data from 364 critically ill patients with acute consciousness impairment (GCS ≤ 11 or FOUR ≤ 12) of various etiologies and without recent signs of seizures from a prospective randomized trial were retrospectively analyzed. Random forest classifiers were trained using 8 visual EEG features-first alone, then in combination with clinical features-to predict survival at 6 months or favorable functional outcome (defined as cerebral performance category 1-2). RESULTS: The area under the ROC curve was 0.812 for predicting survival and 0.790 for predicting favorable outcome using EEG features. Adding clinical features did not improve the overall performance of the classifier (for survival: AUC = 0.806, p = 0.926; for favorable outcome: AUC = 0.777, p = 0.844). Survival could be predicted in all etiology groups: the AUC was 0.958 for patients with HIE, 0.955 for patients with TBI and other neurosurgical diagnoses, 0.697 for patients with metabolic, inflammatory or infectious causes for consciousness impairment and 0.695 for patients with stroke. Training the classifier separately on subgroups of patients with a given etiology (and thus using less training data) leads to poorer classification performance. CONCLUSIONS: While prognostication was best for patients with HIE and TBI, our study demonstrates that similar EEG criteria can be used in patients with various causes of consciousness impairment, and that the size of the training set is more important than homogeneity of ACI etiology.

EEG-based outcome prediction after cardiac arrest with convolutional neural networks: Performance and visualization of discriminative features.

Prognostication for comatose patients after cardiac arrest is a difficult but essential task. Currently, visual interpretation of electroencephalogram (EEG) is one of the main modality used in outcome prediction. There is a growing interest in computer-assisted EEG interpretation, either to overcome the possible subjectivity of visual interpretation, or to identify complex features of the EEG signal. We used a one-dimensional convolutional neural network (CNN) to predict functional outcome based on 19-channel-EEG recorded from 267 adult comatose patients during targeted temperature management after CA. The area under the receiver operating characteristic curve (AUC) on the test set was 0.885. Interestingly, model architecture and fine-tuning only played a marginal role in classification performance. We then used gradient-weighted class activation mapping (Grad-CAM) as visualization technique to identify which EEG features were used by the network to classify an EEG epoch as favorable or unfavorable outcome, and also to understand failures of the network. Grad-CAM showed that the network relied on similar features than classical visual analysis for predicting unfavorable outcome (suppressed background, epileptiform transients). This study confirms that CNNs are promising models for EEG-based prognostication in comatose patients, and that Grad-CAM can provide explanation for the models' decision-making, which is of utmost importance for future use of deep learning models in a clinical setting.

Chow-Liu trees are sufficient predictive models for reproducing key features of functional networks of periictal EEG time-series.

Seizure freedom in patients suffering from pharmacoresistant epilepsies is still not achieved in 20-30% of all cases. Hence, current therapies need to be improved, based on a more complete understanding of ictogenesis. In this respect, the analysis of functional networks derived from intracranial electroencephalographic (iEEG) data has recently become a standard tool. Functional networks however are purely descriptive models and thus are conceptually unable to predict fundamental features of iEEG time-series, e.g., in the context of therapeutical brain stimulation. In this paper we present some first steps towards overcoming the limitations of functional network analysis, by showing that its results are implied by a simple predictive model of time-sliced iEEG time-series. More specifically, we learn distinct graphical models (so called Chow-Liu (CL) trees) as models for the spatial dependencies between iEEG signals. Bayesian inference is then applied to the CL trees, allowing for an analytic derivation/prediction of functional networks, based on thresholding of the absolute value Pearson correlation coefficient (CC) matrix. Using various measures, the thus obtained networks are then compared to those which were derived in the classical way from the empirical CC-matrix. In the high threshold limit we find (a) an excellent agreement between the two networks and (b) key features of periictal networks as they have previously been reported in the literature. Apart from functional networks, both matrices are also compared element-wise, showing that the CL approach leads to a sparse representation, by setting small correlations to values close to zero while preserving the larger ones. Overall, this paper shows the validity of CL-trees as simple, spatially predictive models for periictal iEEG data. Moreover, we suggest straightforward generalizations of the CL-approach for modeling also the temporal features of iEEG signals.

Developmental self-construction and -configuration of functional neocortical neuronal networks.

The prenatal development of neural circuits must provide sufficient configuration to support at least a set of core postnatal behaviors. Although knowledge of various genetic and cellular aspects of development is accumulating rapidly, there is less systematic understanding of how these various processes play together in order to construct such functional networks. Here we make some steps toward such understanding by demonstrating through detailed simulations how a competitive co-operative ('winner-take-all', WTA) network architecture can arise by development from a single precursor cell. This precursor is granted a simplified gene regulatory network that directs cell mitosis, differentiation, migration, neurite outgrowth and synaptogenesis. Once initial axonal connection patterns are established, their synaptic weights undergo homeostatic unsupervised learning that is shaped by wave-like input patterns. We demonstrate how this autonomous genetically directed developmental sequence can give rise to self-calibrated WTA networks, and compare our simulation results with biological data.

Detecting functional hubs of ictogenic networks.

Quantitative EEG (qEEG) has modified our understanding of epileptic seizures, shifting our view from the traditionally accepted hyper-synchrony paradigm toward more complex models based on re-organization of functional networks. However, qEEG measurements are so far rarely considered during the clinical decision-making process. To better understand the dynamics of intracranial EEG signals, we examine a functional network derived from the quantification of information flow between intracranial EEG signals. Using transfer entropy, we analyzed 198 seizures from 27 patients undergoing pre-surgical evaluation for pharmaco-resistant epilepsy. During each seizure we considered for each network the in-, out- and total "hubs", defined respectively as the time and the EEG channels with the maximal incoming, outgoing or total (bidirectional) information flow. In the majority of cases we found that the hubs occur around the middle of seizures, and interestingly not at the beginning or end, where the most dramatic EEG signal changes are found by visual inspection. For the patients who then underwent surgery, good postoperative clinical outcome was on average associated with a higher percentage of out- or total-hubs located in the resected area (for out-hubs p = 0.01, for total-hubs p = 0.04). The location of in-hubs showed no clear predictive value. We conclude that the study of functional networks based on qEEG measurements may help to identify brain areas that are critical for seizure generation and are thus potential targets for focused therapeutic interventions.

Developmental origin of patchy axonal connectivity in the neocortex: a computational model.

Injections of neural tracers into many mammalian neocortical areas reveal a common patchy motif of clustered axonal projections. We studied in simulation a mathematical model for neuronal development in order to investigate how this patchy connectivity could arise in layer II/III of the neocortex. In our model, individual neurons of this layer expressed the activator-inhibitor components of a Gierer-Meinhardt reaction-diffusion system. The resultant steady-state reaction-diffusion pattern across the neuronal population was approximately hexagonal. Growth cones at the tips of extending axons used the various morphogens secreted by intrapatch neurons as guidance cues to direct their growth and invoke axonal arborization, so yielding a patchy distribution of arborization across the entire layer II/III. We found that adjustment of a single parameter yields the intriguing linear relationship between average patch diameter and interpatch spacing that has been observed experimentally over many cortical areas and species. We conclude that a simple Gierer-Meinhardt system expressed by the neurons of the developing neocortex is sufficient to explain the patterns of clustered connectivity observed experimentally.

Simulating cortical development as a self constructing process: a novel multi-scale approach combining molecular and physical aspects.

Current models of embryological development focus on intracellular processes such as gene expression and protein networks, rather than on the complex relationship between subcellular processes and the collective cellular organization these processes support. We have explored this collective behavior in the context of neocortical development, by modeling the expansion of a small number of progenitor cells into a laminated cortex with layer and cell type specific projections. The developmental process is steered by a formal language analogous to genomic instructions, and takes place in a physically realistic three-dimensional environment. A common genome inserted into individual cells control their individual behaviors, and thereby gives rise to collective developmental sequences in a biologically plausible manner. The simulation begins with a single progenitor cell containing the artificial genome. This progenitor then gives rise through a lineage of offspring to distinct populations of neuronal precursors that migrate to form the cortical laminae. The precursors differentiate by extending dendrites and axons, which reproduce the experimentally determined branching patterns of a number of different neuronal cell types observed in the cat visual cortex. This result is the first comprehensive demonstration of the principles of self-construction whereby the cortical architecture develops. In addition, our model makes several testable predictions concerning cell migration and branching mechanisms.

The predictive value of highly malignant EEG patterns after cardiac arrest: evaluation of the ERC-ESICM recommendations.

PURPOSE: The 2021 guidelines endorsed by the European Resuscitation Council (ERC) and the European Society of Intensive Care Medicine (ESICM) recommend using highly malignant electroencephalogram (EEG) patterns (HMEP; suppression or burst-suppression) at > 24 h after cardiac arrest (CA) in combination with at least one other concordant predictor to prognosticate poor neurological outcome. We evaluated the prognostic accuracy of HMEP in a large multicentre cohort and investigated the added value of absent EEG reactivity. METHODS: This is a pre-planned prognostic substudy of the Targeted Temperature Management trial 2. The presence of HMEP and background reactivity to external stimuli on EEG recorded > 24 h after CA was prospectively reported. Poor outcome was measured at 6 months and defined as a modified Rankin Scale score of 4-6. Prognostication was multimodal, and withdrawal of life-sustaining therapy (WLST) was not allowed before 96 h after CA. RESULTS: 845 patients at 59 sites were included. Of these, 579 (69%) had poor outcome, including 304 (36%) with WLST due to poor neurological prognosis. EEG was recorded at a median of 71 h (interquartile range [IQR] 52-93) after CA. HMEP at > 24 h from CA had 50% [95% confidence interval [CI] 46-54] sensitivity and 93% [90-96] specificity to predict poor outcome. Specificity was similar (93%) in 541 patients without WLST. When HMEP were unreactive, specificity improved to 97% [94-99] (p = 0.008). CONCLUSION: The specificity of the ERC-ESICM-recommended EEG patterns for predicting poor outcome after CA exceeds 90% but is lower than in previous studies, suggesting that large-scale implementation may reduce their accuracy. Combining HMEP with an unreactive EEG background significantly improved specificity. As in other prognostication studies, a self-fulfilling prophecy bias may have contributed to observed results.

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.

The effect of sedation and time after cardiac arrest on coma outcome prognostication based on EEG power spectra.

Early prognostication of long-term outcome of comatose patients after cardiac arrest remains challenging. Electroencephalography-based power spectra after cardiac arrest have been shown to help with the identification of patients with favourable outcome during the first day of coma. Here, we aim at comparing the power spectra prognostic value during the first and second day after coma onset following cardiac arrest and to investigate the impact of sedation on prognostication. In this cohort observational study, we included comatose patients (N = 91) after cardiac arrest for whom resting-state electroencephalography was collected on the first and second day after cardiac arrest in four Swiss hospitals. We evaluated whether the average power spectra values at 4.6-15.2 Hz were predictive of patients' outcome based on the best cerebral performance category score at 3 months, with scores ranging from 1 to 5 and dichotomized as favourable (1-2) and unfavourable (3-5). We assessed the effect of sedation and its interaction with the electroencephalography-based power spectra on patient outcome prediction through a generalized linear mixed model. Power spectra values provided 100% positive predictive value (95% confidence intervals: 0.81-1.00) on the first day of coma, with correctly predicted 18 out of 45 favourable outcome patients. On the second day, power spectra values were not predictive of patients' outcome (positive predictive value: 0.46, 95% confidence intervals: 0.19-0.75). On the first day, we did not find evidence of any significant contribution of sedative infusion rates to the patient outcome prediction (P > 0.05). Comatose patients' outcome prediction based on electroencephalographic power spectra is higher on the first compared with the second day after cardiac arrest. Sedation does not appear to impact patient outcome prediction.

Frequency and evolution of sleep-wake disturbances after ischemic stroke: A 2-year prospective study of 437 patients.

OBJECTIVE: In the absence of systematic and longitudinal data, this study prospectively assessed both frequency and evolution of sleep-wake disturbances (SWD) after stroke. METHODS: In 437 consecutively recruited patients with ischemic stroke or transient ischemic attack (TIA), stroke characteristics and outcome were assessed within the 1st week and 3.2 ± 0.3 years (M±SD) after the acute event. SWD were assessed by interview and questionnaires at 1 and 3 months as well as 1 and 2 years after the acute event. Sleep disordered breathing (SDB) was assessed by respirography in the acute phase and repeated in one fifth of the participants 3 months and 1 year later. RESULTS: Patients (63.8% male, 87% ischemic stroke and mean age 65.1 ± 13.0 years) presented with mean NIHSS-score of 3.5 ± 4.5 at admission. In the acute phase, respiratory event index was >15/h in 34% and >30/h in 15% of patients. Over the entire observation period, the frequencies of excessive daytime sleepiness (EDS), fatigue and insomnia varied between 10-14%, 22-28% and 20-28%, respectively. Mean insomnia and EDS scores decreased from acute to chronic stroke, whereas restless legs syndrome (RLS) percentages (6-9%) and mean fatigue scores remained similar. Mean self-reported sleep duration was enhanced at acute stroke (month 1: 07:54 ± 01:27h) and decreased at chronic stage (year 2: 07:43 ± 01:20h). CONCLUSIONS: This study documents a high frequency of SDB, insomnia, fatigue and a prolonged sleep duration after stroke/TIA, which can persist for years. Considering the negative effects of SWD on physical, brain and mental health these data suggest the need for a systematic assessment and management of post-stroke SWD.

Isolated mandibular sleep-related rhythmic movement disorder: A case report.

Background: Rhythmic masticatory muscle activity (RMMA) in sleep is usually not considered pathological unless associated with bruxism. On the other hand, so-called sleep-related rhythmic movement disorders (SRRMD) are a recognized category of sleep disorders, which involve prolonged rhythmic activity of large muscle groups, such as the whole body, the head, or a limb, but typically not the masticatory muscles.Clinical Presentation: A polysomnographic description of a patient with symptomatic RMMA without bruxism, fulfilling the diagnostic criteria of an SRRMD, is presented. The symptoms were initially misdiagnosed as bruxism and then as sleep-related epilepsy, which delayed an adequate treatment. Therapy of the comorbid obstructive sleep apnea with a positive airway pressure device (APAP) led to a self-reported improvement.Conclusion: The differential diagnosis of jaw movement in sleep is vast; a correct diagnosis is of the essence for adequate treatment. The prevalence of isolated RMMA resulting in perturbation of sleep warrants further exploration.

Heart rate complexity: An early prognostic marker of patient outcome after cardiac arrest.

OBJECTIVE: Early prognostication in comatose patients after cardiac arrest (CA) is difficult but essential to inform relatives and optimize treatment. Here we investigate the predictive value of heart-rate variability captured by multiscale entropy (MSE) for long-term outcomes in comatose patients during the first 24 hours after CA. METHODS: In this retrospective analysis of prospective multi-centric cohort, we analyzed MSE of the heart rate in 79 comatose patients after CA while undergoing targeted temperature management and sedation during the first day of coma. From the MSE, two complexity indices were derived by summing values over short and long time scales (CIs and CIl). We splitted the data in training and test datasets for analysing the predictive value for patient outcomes (defined as best cerebral performance category within 3 months) of CIs and CIl. RESULTS: Across the whole dataset, CIl provided the best sensitivity, specificity, and accuracy (88%, 75%, and 82%, respectively). Positive and negative predictive power were 81% and 84%. CONCLUSIONS: Characterizing the complexity of the ECG in patients after CA provides an accurate prediction of both favorable and unfavorable outcomes. SIGNIFICANCE: The analysis of heartrate variability by means of MSE provides accurate outcome prediction on the first day of coma.

Diagnostic and prognostic EEG analysis of critically ill patients: A deep learning study.

Visual interpretation of electroencephalography (EEG) is time consuming, may lack objectivity, and is restricted to features detectable by a human. Computer-based approaches, especially deep learning, could potentially overcome these limitations. However, most deep learning studies focus on a specific question or a single pathology. Here we explore the potential of deep learning for EEG-based diagnostic and prognostic assessment of patients with acute consciousness impairment (ACI) of various etiologies. EEGs from 358 adults from a randomized controlled trial (CERTA, NCT03129438) were retrospectively analyzed. A convolutional neural network was used to predict the clinical outcome (based either on survival or on best cerebral performance category) and to determine the etiology (four diagnostic categories). The largest probability output served as marker for the confidence of the network in its prediction ("certainty factor"); we also systematically compared the predictions with raw EEG data, and used a visualization algorithm (Grad-CAM) to highlight discriminative patterns. When all patients were considered, the area under the receiver operating characteristic curve (AUC) was 0.721 for predicting survival and 0.703 for predicting the outcome based on best CPC; for patients with certainty factor ≥ 60 % the AUCs increased to 0.776 and 0.755 respectively; and for certainty factor ≥ 75 % to 0.852 and 0.879. The accuracy for predicting the etiology was 54.5 %; the accuracy increased to 67.7 %, 70.3 % and 84.1 % for patients with certainty factor of 50 %, 60 % and 75 % respectively. Visual analysis showed that the network learnt EEG patterns typically recognized by human experts, and suggested new criteria. This work demonstrates for the first time the potential of deep learning-based EEG analysis in critically ill patients with various etiologies of ACI. Certainty factor and post-hoc correlation of input data with prediction help to better characterize the method and pave the route for future implementations in clinical routine.

Bayesian brain in tinnitus: Computational modeling of three perceptual phenomena using a modified Hierarchical Gaussian Filter.

Recently, Bayesian brain-based models emerged as a possible composite of existing theories, providing an universal explanation of tinnitus phenomena. Yet, the involvement of multiple synergistic mechanisms complicates the identification of behavioral and physiological evidence. To overcome this, an empirically tested computational model could support the evaluation of theoretical hypotheses by intrinsically encompassing different mechanisms. The aim of this work was to develop a generative computational tinnitus perception model based on the Bayesian brain concept. The behavioral responses of 46 tinnitus subjects who underwent ten consecutive residual inhibition assessments were used for model fitting. Our model was able to replicate the behavioral responses during residual inhibition in our cohort (median linear correlation coefficient of 0.79). Using the same model, we simulated two additional tinnitus phenomena: residual excitation and occurrence of tinnitus in non-tinnitus subjects after sensory deprivation. In the simulations, the trajectories of the model were consistent with previously obtained behavioral and physiological observations. Our work introduces generative computational modeling to the research field of tinnitus. It has the potential to quantitatively link experimental observations to theoretical hypotheses and to support the search for neural signatures of tinnitus by finding correlates between the latent variables of the model and measured physiological data.