A Deep Learning Framework for Deriving Noninvasive Intracranial Pressure Waveforms from Transcranial Doppler.

Increased intracranial pressure (ICP) causes disability and mortality in the neurointensive care population. Current methods for monitoring ICP are invasive. We designed a deep learning framework using a domain adversarial neural network to estimate noninvasive ICP, from blood pressure, electrocardiogram, and cerebral blood flow velocity. Our model had a mean of median absolute error of 3.88 ± 3.26 mmHg for the domain adversarial neural network, and 3.94 ± 1.71 mmHg for the domain adversarial transformers. Compared with nonlinear approaches, such as support vector regression, this was 26.7% and 25.7% lower. Our proposed framework provides more accurate noninvasive ICP estimates than currently available. ANN NEUROL 2023;94:196-202.

Injury patterns associated with cognitive motor dissociation.

In unconscious appearing patients with acute brain injury, wilful brain activation to motor commands without behavioural signs of command following, known as cognitive motor dissociation (CMD), is associated with functional recovery. CMD can be detected by applying machine learning to EEG recorded during motor command presentation in behaviourally unresponsive patients. Identifying patients with CMD carries clinical implications for patient interactions, communication with families, and guidance of therapeutic decisions but underlying mechanisms of CMD remain unknown. By analysing structural lesion patterns and network level dysfunction we tested the hypothesis that, in cases with preserved arousal and command comprehension, a failure to integrate comprehended motor commands with motor outputs underlies CMD. Manual segmentation of T2-fluid attenuated inversion recovery and diffusion weighted imaging sequences quantifying structural injury was performed in consecutive unresponsive patients with acute brain injury (n = 107) who underwent EEG-based CMD assessments and MRI. Lesion pattern analysis was applied to identify lesion patterns common among patients with (n = 21) and without CMD (n = 86). Thalamocortical and cortico-cortical network connectivity were assessed applying ABCD classification of power spectral density plots and weighted pairwise phase consistency (WPPC) to resting EEG, respectively. Two distinct structural lesion patterns were identified on MRI for CMD and three for non-CMD patients. In non-CMD patients, injury to brainstem arousal pathways including the midbrain were seen, while no CMD patients had midbrain lesions. A group of non-CMD patients was identified with injury to the left thalamus, implicating possible language comprehension difficulties. Shared lesion patterns of globus pallidus and putamen were seen for a group of CMD patients, which have been implicated as part of the anterior forebrain mesocircuit in patients with reversible disorders of consciousness. Thalamocortical network dysfunction was less common in CMD patients [ABCD-index 2.3 (interquartile range, IQR 2.1-3.0) versus 1.4 (IQR 1.0-2.0), P < 0.0001; presence of D 36% versus 3%, P = 0.0006], but WPPC was not different. Bilateral cortical lesions were seen in patients with and without CMD. Thalamocortical disruption did not differ for those with CMD, but long-range WPPC was decreased in 1-4 Hz [odds ratio (OR) 0.8; 95% confidence interval (CI) 0.7-0.9] and increased in 14-30 Hz frequency ranges (OR 1.2; 95% CI 1.0-1.5). These structural and functional data implicate a failure of motor command integration at the anterior forebrain mesocircuit level with preserved thalamocortical network function for CMD patients with subcortical lesions. Amongst patients with bilateral cortical lesions preserved cortico-cortical network function is associated with CMD detection. These data may allow screening for CMD based on widely available structural MRI and resting EEG.

A Novel Approach to Screen for Somatosensory Evoked Potentials in Critical Care.

BACKGROUND: Somatosensory evoked potentials (SSEPs) help prognostication, particularly in patients with diffuse brain injury. However, use of SSEP is limited in critical care. We propose a novel, low-cost approach allowing acquisition of screening SSEP using widely available intensive care unit (ICU) equipment, specifically a peripheral "train-of-four" stimulator and standard electroencephalograph. METHODS: The median nerve was stimulated using a train-of-four stimulator, and a standard 21-channel electroencephalograph was recorded to generate the screening SSEP. Generation of the SSEP was supported by visual inspection, univariate event-related potentials statistics, and a multivariate support vector machine (SVM) decoding algorithm. This approach was validated in 15 healthy volunteers and validated against standard SSEPs in 10 ICU patients. The ability of this approach to predict poor neurological outcome, defined as death, vegetative state, or severe disability at 6 months, was tested in an additional set of 39 ICU patients. RESULTS: In each of the healthy volunteers, both the univariate and the SVM methods reliably detected SSEP responses. In patients, when compared against the standard SSEP method, the univariate event-related potentials method matched in nine of ten patients (sensitivity = 94%, specificity = 100%), and the SVM had 100% sensitivity and specificity when compared with the standard method. For the 49 ICU patients, we performed both the univariate and the SVM methods: a bilateral absence of short latency responses (n = 8) predicted poor neurological outcome with 0% FPR (sensitivity = 21%, specificity = 100%). CONCLUSIONS: Somatosensory evoked potentials can reliably be recorded using the proposed approach. Given the very good but slightly lower sensitivity of absent SSEPs in the proposed screening approach, confirmation of absent SSEP responses using standard SSEP recordings is advised.

Optimal Cerebral Perfusion Pressure and Brain Tissue Oxygen in Aneurysmal Subarachnoid Hemorrhage.

BACKGROUND: Targeting a cerebral perfusion pressure optimal for cerebral autoregulation (CPPopt) has been gaining more attention to prevent secondary damage after acute neurological injury. Brain tissue oxygenation (PbtO2) can identify insufficient cerebral blood flow and secondary brain injury. Defining the relationship between CPPopt and PbtO2 after aneurysmal subarachnoid hemorrhage may result in (1) mechanistic insights into whether and how CPPopt-based strategies might be beneficial and (2) establishing support for the use of PbtO2 as an adjunctive monitor for adequate or optimal local perfusion. METHODS: We performed a retrospective analysis of a prospectively collected 2-center dataset of patients with aneurysmal subarachnoid hemorrhage with or without later diagnosis of delayed cerebral ischemia (DCI). CPPopt was calculated as the cerebral perfusion pressure (CPP) value corresponding to the lowest pressure reactivity index (moving correlation coefficient of mean arterial and intracranial pressure). The relationship of (hourly) deltaCPP (CPP-CPPopt) and PbtO2 was investigated using natural spline regression analysis. Data after DCI diagnosis were excluded. Brain tissue hypoxia was defined as PbtO2 <20 mmHg. RESULTS: One hundred thirty-one patients were included with a median of 44.0 (interquartile range, 20.8-78.3) hourly CPPopt/PbtO2 datapoints. The regression plot revealed a nonlinear relationship between PbtO2 and deltaCPP (P<0.001) with PbtO2 decrease with deltaCPP <0 mmHg and stable PbtO2 with deltaCPP ≥0mmHg, although there was substantial individual variation. Brain tissue hypoxia (34.6% of all measurements) was more frequent with deltaCPP <0 mmHg. These dynamics were similar in patients with or without DCI. CONCLUSIONS: We found a nonlinear relationship between PbtO2 and deviation of patients' CPP from CPPopt in aneurysmal subarachnoid hemorrhage patients in the pre-DCI period. CPP values below calculated CPPopt were associated with lower PbtO2. Nevertheless, the nature of PbtO2 measurements is complex, and the variability is high. Combined multimodality monitoring with CPP/CPPopt and PbtO2 should be recommended to redefine individual pressure targets (CPP/CPPopt) and retain the option to detect local perfusion deficits during DCI (PbtO2), which cannot be fulfilled by both measurements interchangeably.

Hemoglobin Concentration Impacts Viscoelastic Hemostatic Assays in ICU Admitted Patients.

OBJECTIVES: Low hemoglobin concentration impairs clinical hemostasis across several diseases. It is unclear whether hemoglobin impacts laboratory functional coagulation assessments. We evaluated the relationship of hemoglobin concentration on viscoelastic hemostatic assays in intracerebral hemorrhage (ICH) and perioperative patients admitted to an ICU. DESIGN: Observational cohort study and separate in vitro laboratory study. SETTING: Multicenter tertiary referral ICUs. PATIENTS: Two acute ICH cohorts receiving distinct testing modalities: rotational thromboelastometry (ROTEM) and thromboelastography (TEG), and a third surgical ICU cohort receiving ROTEM were evaluated to assess the generalizability of findings across disease processes and testing platforms. A separate in vitro ROTEM laboratory study was performed utilizing ICH patient blood samples. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Relationships between baseline hemoglobin and ROTEM/TEG results were separately assessed across patient cohorts using Spearman correlations and linear regression models. A separate in vitro study assessed ROTEM tracing changes after serial hemoglobin modifications from ICH patient blood samples. In both our ROTEM (n = 34) and TEG (n = 239) ICH cohorts, hemoglobin concentrations directly correlated with coagulation kinetics (ROTEM r: 0.46; p = 0.01; TEG r: 0.49; p < 0.0001) and inversely correlated with clot strength (ROTEM r: -0.52, p = 0.002; TEG r: -0.40, p < 0.0001). Similar relationships were identified in perioperative ICU admitted patients (n = 121). We continued to identify these relationships in linear regression models. When manipulating ICH patient blood samples to achieve lower hemoglobin concentrations in vitro, we similarly identified that lower hemoglobin concentrations resulted in progressively faster coagulation kinetics and greater clot strength on ROTEM tracings. CONCLUSIONS: Lower hemoglobin concentrations have a consistent, measurable impact on ROTEM/TEG testing in ICU admitted patients, which appear to be artifactual. It is possible that patients with low hemoglobin may appear to have normal viscoelastic parameters when, in fact, they have a mild hypocoagulable state. Further work is required to determine if these tests should be corrected for a patient's hemoglobin concentration.

Prolonged Unconsciousness is Common in COVID-19 and Associated with Hypoxemia.

OBJECTIVE: The purpose of this study was to estimate the time to recovery of command-following and associations between hypoxemia with time to recovery of command-following. METHODS: In this multicenter, retrospective, cohort study during the initial surge of the United States' pandemic (March-July 2020) we estimate the time from intubation to recovery of command-following, using Kaplan Meier cumulative-incidence curves and Cox proportional hazard models. Patients were included if they were admitted to 1 of 3 hospitals because of severe coronavirus disease 2019 (COVID-19), required endotracheal intubation for at least 7 days, and experienced impairment of consciousness (Glasgow Coma Scale motor score <6). RESULTS: Five hundred seventy-one patients of the 795 patients recovered command-following. The median time to recovery of command-following was 30 days (95% confidence interval [CI] = 27-32 days). Median time to recovery of command-following increased by 16 days for patients with at least one episode of an arterial partial pressure of oxygen (PaO2 ) value ≤55 mmHg (p < 0.001), and 25% recovered ≥10 days after cessation of mechanical ventilation. The time to recovery of command-following  was associated with hypoxemia (PaO2 ≤55 mmHg hazard ratio [HR] = 0.56, 95% CI = 0.46-0.68; PaO2 ≤70 HR = 0.88, 95% CI = 0.85-0.91), and each additional day of hypoxemia decreased the likelihood of recovery, accounting for confounders including sedation. These findings were confirmed among patients without any imagining evidence of structural brain injury (n = 199), and in a non-overlapping second surge cohort (N = 427, October 2020 to April 2021). INTERPRETATION: Survivors of severe COVID-19 commonly recover consciousness weeks after cessation of mechanical ventilation. Long recovery periods are associated with more severe hypoxemia. This relationship is not explained by sedation or brain injury identified on clinical imaging and should inform decisions about life-sustaining therapies. ANN NEUROL 2022;91:740-755.

Classification of Level of Consciousness in a Neurological ICU Using Physiological Data.

BACKGROUND: Impaired consciousness is common in intensive care unit (ICU) patients, and an individual's degree of consciousness is crucial to determining their care and prognosis. However, there are no methods that continuously monitor consciousness and alert clinicians to changes. We investigated the use of physiological signals collected in the ICU to classify levels of consciousness in critically ill patients. METHODS: We studied 61 patients with subarachnoid hemorrhage (SAH) and 178 patients with intracerebral hemorrhage (ICH) from the neurological ICU at Columbia University Medical Center in a retrospective observational study of prospectively collected data. The level of consciousness was determined on the basis of neurological examination and mapped to comatose, vegetative state or unresponsive wakefulness syndrome (VS/UWS), minimally conscious minus state (MCS-), and command following. For each physiological signal, we extracted time-series features and performed classification using extreme gradient boosting on multiple clinically relevant tasks across subsets of physiological signals. We applied this approach independently on both SAH and ICH patient groups for three sets of variables: (1) a minimal set common to most hospital patients (e.g., heart rate), (2) variables available in most ICUs (e.g., body temperature), and (3) an extended set recorded mainly in neurological ICUs (absent for the ICH patient group; e.g., brain temperature). RESULTS: On the commonly performed classification task of VS/UWS versus MCS-, we achieved an area under the receiver operating characteristic curve (AUROC) in the SAH patient group of 0.72 (sensitivity 82%, specificity 57%; 95% confidence interval [CI] 0.63-0.81) using the extended set, 0.69 (sensitivity 83%, specificity 51%; 95% CI 0.59-0.78) on the variable set available in most ICUs, and 0.69 (sensitivity 56%, specificity 78%; 95% CI 0.60-0.78) on the minimal set. In the ICH patient group, AUROC was 0.64 (sensitivity 56%, specificity 65%; 95% CI 0.55-0.74) using the minimal set and 0.61 (sensitivity 50%, specificity 80%; 95% CI 0.51-0.71) using the variables available in most ICUs. CONCLUSIONS: We find that physiological signals can be used to classify states of consciousness for patients in the ICU. Building on this with intraday assessments and increasing sensitivity and specificity may enable alarm systems that alert physicians to changes in consciousness and frequent monitoring of consciousness throughout the day, both of which may improve patient care and outcomes.

Suboptimal Cerebral Perfusion is Associated with Ischemia After Intracerebral Hemorrhage.

BACKGROUND: Remote ischemic lesions on diffusion-weighted imaging (DWI) occur in one third of patients with intracerebral hemorrhage (ICH) and are associated with worse outcomes. The etiology is unclear and not solely due to blood pressure reduction. We hypothesized that impaired cerebrovascular autoregulation and hypoperfusion below individualized lower limits of autoregulation are associated with the presence of DWI lesions. METHODS: This was a retrospective, single-center study of all primary ICH with intraparenchymal pressure monitoring within 10 days from onset and subsequent magnetic resonance imaging. Pressure reactivity index was calculated as the correlation coefficient between mean arterial pressure and intracranial pressure. Optimal cerebral perfusion pressure (CPPopt) is the cerebral perfusion pressure (CPP) with the lowest corresponding pressure reactivity index. The difference between CPP and CPPopt, time spent below the lower limit of autoregulation (LLA), and time spent above the upper limit of autoregulation (ULA) were calculated by using mean hourly physiologic data. Univariate associations between physiologic parameters and DWI lesions were analyzed by using binary logistic regression. RESULTS: A total of 505 h of artifact-free data from seven patients without DWI lesions and 479 h from six patients with DWI lesions were analyzed. Patients with DWI lesions had higher intracranial pressure (17.50 vs. 10.92 mm Hg; odds ratio 1.14, confidence interval 1.01-1.29) but no difference in mean arterial pressure or CPP compared with patients without DWI lesions. The presence of DWI lesions was significantly associated with a greater percentage of time spent below the LLA (49.85% vs. 14.70%, odds ratio 5.77, confidence interval 1.88-17.75). No significant association was demonstrated between CPPopt, the difference between CPP and CPPopt, ULA, LLA, or time spent above the ULA between groups. CONCLUSIONS: Blood pressure reduction below the LLA is associated with ischemia after acute ICH. Individualized, autoregulation-informed targets for blood pressure reduction may provide a novel paradigm in acute management of ICH and require further study.

Long-Term Outcomes of Patients with Stroke Predicted by Clinicians to have no Chance of Meaningful Recovery: A Japanese Cohort Study.

BACKGROUND: Little is known about the natural history of comatose patients with brain injury, as in many countries most of these patients die in the context of withdrawal of life-sustaining therapies (WLSTs). The accuracy of predicting recovery that is used to guide goals-of-care decisions is uncertain. We examined long-term outcomes of patients with ischemic or hemorrhagic stroke predicted by experienced clinicians to have no chance of meaningful recovery in Japan, where WLST in patients with isolated neurological disease is uncommon. METHODS: We retrospectively reviewed the medical records of all patients admitted with acute ischemic stroke, intracerebral hemorrhage, or nontraumatic subarachnoid hemorrhage between January 2018 and December 2020 to a neurocritical care unit at Toda Medical Group Asaka Medical Center in Saitama, Japan. We screened for patients who were predicted by the attending physician on postinjury day 1-4 to have no chance of meaningful recovery. Primary outcome measures were disposition at hospital discharge and the ability to follow commands and functional outcomes measured by the Glasgow Outcome Scale-Extended (GOS-E), which was assessed 6 months after injury. RESULTS: From 860 screened patients, we identified 40 patients (14 with acute ischemic stroke, 19 with intracerebral hemorrhage, and 7 with subarachnoid hemorrhage) who were predicted to have no chance of meaningful recovery. Median age was 77 years (interquartile range 64-85), 53% (n = 21) were women, and 80% (n = 32) had no functional deficits prior to hospitalization. Six months after injury, 17 patients were dead, 14 lived in a long-term care hospital, 3 lived at home, 2 lived in a rehabilitation center, and 2 lived in a nursing home. Three patients reliably followed commands, two were in a vegetative state (GOS-E 2), four fully depended on others and required constant assistance (GOS-E 3), one could be left alone independently for 8 h per day but remained dependent (GOS-E 4), and one was independent and able to return to work-like activities (GOS-E 5). CONCLUSIONS: In the absence of WLST, almost half of the patients predicted shortly after the injury to have no chance of meaningful recovery were dead 6 months after the injury. A small minority of patients had good functional recovery, highlighting the need for more accurate neurological prognostication.

Detection of Brain Activation in Unresponsive Patients with Acute Brain Injury.

BACKGROUND: Brain activation in response to spoken motor commands can be detected by electroencephalography (EEG) in clinically unresponsive patients. The prevalence and prognostic importance of a dissociation between commanded motor behavior and brain activation in the first few days after brain injury are not well understood. METHODS: We studied a prospective, consecutive series of patients in a single intensive care unit who had acute brain injury from a variety of causes and who were unresponsive to spoken commands, including some patients with the ability to localize painful stimuli or to fixate on or track visual stimuli. Machine learning was applied to EEG recordings to detect brain activation in response to commands that patients move their hands. The functional outcome at 12 months was determined with the Glasgow Outcome Scale-Extended (GOS-E; levels range from 1 to 8, with higher levels indicating better outcomes). RESULTS: A total of 16 of 104 unresponsive patients (15%) had brain activation detected by EEG at a median of 4 days after injury. The condition in 8 of these 16 patients (50%) and in 23 of 88 patients (26%) without brain activation improved such that they were able to follow commands before discharge. At 12 months, 7 of 16 patients (44%) with brain activation and 12 of 84 patients (14%) without brain activation had a GOS-E level of 4 or higher, denoting the ability to function independently for 8 hours (odds ratio, 4.6; 95% confidence interval, 1.2 to 17.1). CONCLUSIONS: A dissociation between the absence of behavioral responses to motor commands and the evidence of brain activation in response to these commands in EEG recordings was found in 15% of patients in a consecutive series of patients with acute brain injury. (Supported by the Dana Foundation and the James S. McDonnell Foundation.).

Vector Angle Analysis of Multimodal Neuromonitoring Data for Continuous Prediction of Delayed Cerebral Ischemia.

BACKGROUND: Dysfunctional cerebral autoregulation often precedes delayed cerebral ischemia (DCI). Currently, there are no data-driven techniques that leverage this information to predict DCI in real time. Our hypothesis is that information using continuous updated analyses of multimodal neuromonitoring and cerebral autoregulation can be deployed to predict DCI. METHODS: Time series values of intracranial pressure, brain tissue oxygenation, cerebral perfusion pressure (CPP), optimal CPP (CPPOpt), ΔCPP (CPP - CPPOpt), mean arterial pressure, and pressure reactivity index were combined and summarized as vectors. A validated temporal signal angle measurement was modified into a classification algorithm that incorporates hourly data. The time-varying temporal signal angle measurement (TTSAM) algorithm classifies DCI at varying time points by vectorizing and computing the angle between the test and reference time signals. The patient is classified as DCI+ if the error between the time-varying test vector and DCI+ reference vector is smaller than that between the time-varying test vector and DCI- reference vector. Finally, prediction at time point t is calculated as the majority voting over all the available signals. The leave-one-patient-out cross-validation technique was used to train and report the performance of the algorithms. The TTSAM and classifier performance was determined by balanced accuracy, F1 score, true positive, true negative, false positive, and false negative over time. RESULTS: One hundred thirty-one patients with aneurysmal subarachnoid hemorrhage who underwent multimodal neuromonitoring were identified from two centers (Columbia University: 52 [39.7%], Aachen University: 79 [60.3%]) and included in the analysis. Sixty-four (48.5%) patients had DCI, and DCI was diagnosed 7.2 ± 3.3 days after hemorrhage. The TTSAM algorithm achieved a balanced accuracy of 67.3% and an F1 score of 0.68 at 165 h (6.9 days) from bleed day with a true positive of 0.83, false positive of 0.16, true negative of 0.51, and false negative of 0.49. CONCLUSIONS: A TTSAM algorithm using multimodal neuromonitoring and cerebral autoregulation calculations shows promise to classify DCI in real time.

Predicting Shunt Dependency from the Effect of Cerebrospinal Fluid Drainage on Ventricular Size.

BACKGROUND: Prolonged external ventricular drainage (EVD) in patients with subarachnoid hemorrhage (SAH) leads to morbidity, whereas early removal can have untoward effects related to recurrent hydrocephalus. A metric to help determine the optimal time for EVD removal or ventriculoperitoneal shunt (VPS) placement would be beneficial in preventing the prolonged, unnecessary use of EVD. This study aimed to identify whether dynamics of cerebrospinal fluid (CSF) biometrics can temporally predict VPS dependency after SAH. METHODS: This was a retrospective analysis of a prospective, single-center, observational study of patients with aneurysmal SAH who required EVD placement for hydrocephalus. Patients were divided into VPS-dependent (VPS+) and non-VPS dependent groups. We measured the bicaudate index (BCI) on all available computed tomography scans and calculated the change over time (ΔBCI). We analyzed the relationship of ΔBCI with CSF output by using Pearson's correlation. A k-nearest neighbor model of the relationship between ΔBCI and CSF output was computed to classify VPS. RESULTS: Fifty-eight patients met inclusion criteria. CSF output was significantly higher in the VPS+ group in the 7 days post EVD placement. There was a negative correlation between delta BCI and CSF output in the VPS+ group (negative delta BCI means ventricles become smaller) and a positive correlation in the VPS- group starting from days four to six after EVD placement (p < 0.05). A weighted k-nearest neighbor model for classification had a sensitivity of 0.75, a specificity of 0.70, and an area under the receiver operating characteristic curve of 0.80. CONCLUSIONS: The correlation of ΔBCI and CSF output is a reliable intraindividual biometric for VPS dependency after SAH as early as days four to six after EVD placement. Our machine learning model leverages this relationship between ΔBCI and cumulative CSF output to predict VPS dependency. Early knowledge of VPS dependency could be studied to reduce EVD duration in many centers (intensive care unit length of stay).

Dynamic Intracranial Pressure Waveform Morphology Predicts Ventriculitis.

BACKGROUND: Intracranial pressure waveform morphology reflects compliance, which can be decreased by ventriculitis. We investigated whether morphologic analysis of intracranial pressure dynamics predicts the onset of ventriculitis. METHODS: Ventriculitis was defined as culture or Gram stain positive cerebrospinal fluid, warranting treatment. We developed a pipeline to automatically isolate segments of intracranial pressure waveforms from extraventricular catheters, extract dominant pulses, and obtain morphologically similar groupings. We used a previously validated clinician-supervised active learning paradigm to identify metaclusters of triphasic, single-peak, or artifactual peaks. Metacluster distributions were concatenated with temperature and routine blood laboratory values to create feature vectors. A L2-regularized logistic regression classifier was trained to distinguish patients with ventriculitis from matched controls, and the discriminative performance using area under receiver operating characteristic curve with bootstrapping cross-validation was reported. RESULTS: Fifty-eight patients were included for analysis. Twenty-seven patients with ventriculitis from two centers were identified. Thirty-one patients with catheters but without ventriculitis were selected as matched controls based on age, sex, and primary diagnosis. There were 1590 h of segmented data, including 396,130 dominant pulses in patients with ventriculitis and 557,435 pulses in patients without ventriculitis. There were significant differences in metacluster distribution comparing before culture-positivity versus during culture-positivity (p < 0.001) and after culture-positivity (p < 0.001). The classifier demonstrated good discrimination with median area under receiver operating characteristic 0.70 (interquartile range 0.55-0.80). There were 1.5 true alerts (ventriculitis detected) for every false alert. CONCLUSIONS: Intracranial pressure waveform morphology analysis can classify ventriculitis without cerebrospinal fluid sampling.

Disorders of Consciousness in Hospitalized Patients with COVID-19: The Role of the Systemic Inflammatory Response Syndrome.

BACKGROUND: Prevalence and etiology of unconsciousness are uncertain in hospitalized patients with coronavirus disease 2019 (COVID-19). We tested the hypothesis that increased inflammation in COVID-19 precedes coma, independent of medications, hypotension, and hypoxia. METHODS: We retrospectively assessed 3203 hospitalized patients with COVID-19 from March 2 through July 30, 2020, in New York City with the Glasgow Coma Scale and systemic inflammatory response syndrome (SIRS) scores. We applied hazard ratio (HR) modeling and mediation analysis to determine the risk of SIRS score elevation to precede coma, accounting for confounders. RESULTS: We obtained behavioral assessments in 3203 of 10,797 patients admitted to the hospital who tested positive for SARS-CoV-2. Of those patients, 1054 (32.9%) were comatose, which first developed on median hospital day 2 (interquartile range [IQR] 1-9). During their hospital stay, 1538 (48%) had a SIRS score of 2 or above at least once, and the median maximum SIRS score was 2 (IQR 1-2). A fivefold increased risk of coma (HR 5.05, 95% confidence interval 4.27-5.98) was seen for each day that patients with COVID-19 had elevated SIRS scores, independent of medication effects, hypotension, and hypoxia. The overall mortality in this population was 13.8% (n = 441). Coma was associated with death (odds ratio 7.77, 95% confidence interval 6.29-9.65) and increased length of stay (13 days [IQR 11.9-14.1] vs. 11 [IQR 9.6-12.4]), accounting for demographics. CONCLUSIONS: Disorders of consciousness are common in hospitalized patients with severe COVID-19 and are associated with increased mortality and length of hospitalization. The underlying etiology of disorders of consciousness in this population is uncertain but, in addition to medication effects, may in part be linked to systemic inflammation.

Dynamic Detection of Delayed Cerebral Ischemia: A Study in 3 Centers.

BACKGROUND AND PURPOSE: Delayed cerebral ischemia (DCI) after aneurysmal subarachnoid hemorrhage negatively impacts long-term recovery but is often detected too late to prevent damage. We aim to develop hourly risk scores using routinely collected clinical data to detect DCI. METHODS: A DCI classification model was trained using vital sign measurements (heart rate, blood pressure, respiratory rate, and oxygen saturation) and demographics routinely collected for clinical care. Twenty-two time-varying physiological measures were computed including mean, SD, and cross-correlation of heart rate time series with each of the other vitals. Classification was achieved using an ensemble approach with L2-regularized logistic regression, random forest, and support vector machines models. Classifier performance was determined by area under the receiver operating characteristic curves and confusion matrices. Hourly DCI risk scores were generated as the posterior probability at time t using the Ensemble classifier on cohorts recruited at 2 external institutions (n=38 and 40). RESULTS: Three hundred ten patients were included in the training model (median, 54 years old [interquartile range, 45-65]; 80.2% women, 28.4% Hunt and Hess scale 4-5, 38.7% Modified Fisher Scale 3-4); 101 (33%) developed DCI with a median onset day 6 (interquartile range, 5-8). Classification accuracy before DCI onset was 0.83 (interquartile range, 0.76-0.83) area under the receiver operating characteristic curve. Risk scores applied to external institution datasets correctly predicted 64% and 91% of DCI events as early as 12 hours before clinical detection, with 2.7 and 1.6 true alerts for every false alert. CONCLUSIONS: An hourly risk score for DCI derived from routine vital signs may have the potential to alert clinicians to DCI, which could reduce neurological injury.

Electrocerebral Signature of Cardiac Death.

BACKGROUND: Electroencephalography (EEG) findings following cardiovascular collapse in death are uncertain. We aimed to characterize EEG changes immediately preceding and following cardiac death. METHODS: We retrospectively analyzed EEGs of patients who died from cardiac arrest while undergoing standard EEG monitoring in an intensive care unit. Patients with brain death preceding cardiac death were excluded. Three events during fatal cardiovascular failure were investigated: (1) last recorded QRS complex on electrocardiogram (QRS0), (2) cessation of cerebral blood flow (CBF0) estimated as the time that blood pressure and heart rate dropped below set thresholds, and (3) electrocerebral silence on EEG (EEG0). We evaluated EEG spectral power, coherence, and permutation entropy at these time points. RESULTS: Among 19 patients who died while undergoing EEG monitoring, seven (37%) had a comfort-measures-only status and 18 (95%) had a do-not-resuscitate status in place at the time of death. EEG0 occurred at the time of QRS0 in five patients and after QRS0 in two patients (cohort median - 2.0, interquartile range - 8.0 to 0.0), whereas EEG0 was seen at the time of CBF0 in six patients and following CBF0 in 11 patients (cohort median 2.0 min, interquartile range - 1.5 to 6.0). After CBF0, full-spectrum log power (p < 0.001) and coherence (p < 0.001) decreased on EEG, whereas delta (p = 0.007) and theta (p < 0.001) permutation entropy increased. CONCLUSIONS: Rarely may patients have transient electrocerebral activity following the last recorded QRS (less than 5 min) and estimated cessation of cerebral blood flow. These results may have implications for discussions around cardiopulmonary resuscitation and organ donation.

Heart Rate Variability as a Biomarker of Neurocardiogenic Injury After Subarachnoid Hemorrhage.

BACKGROUND: The objective of this study was to examine whether heart rate variability (HRV) measures can be used to detect neurocardiogenic injury (NCI). METHODS: Three hundred and twenty-six consecutive admissions with aneurysmal subarachnoid hemorrhage (SAH) met criteria for the study. Of 326 subjects, 56 (17.2%) developed NCI which we defined by wall motion abnormality with ventricular dysfunction on transthoracic echocardiogram or cardiac troponin-I > 0.3 ng/mL without electrocardiogram evidence of coronary artery insufficiency. HRV measures (in time and frequency domains, as well as nonlinear technique of detrended fluctuation analysis) were calculated over the first 48 h. We applied longitudinal multilevel linear regression to characterize the relationship of HRV measures with NCI and examine between-group differences at baseline and over time. RESULTS: There was decreased vagal activity in NCI subjects with a between-group difference in low/high frequency ratio (ß 3.42, SE 0.92, p = 0.0002), with sympathovagal balance in favor of sympathetic nervous activity. All time-domain measures were decreased in SAH subjects with NCI. An ensemble machine learning approach translated these measures into a classification tool that demonstrated good discrimination using the area under the receiver operating characteristic curve (AUROC 0.82), the area under precision recall curve (AUPRC 0.75), and a correct classification rate of 0.81. CONCLUSIONS: HRV measures are significantly associated with our label of NCI and a machine learning approach using features derived from HRV measures can classify SAH patients that develop NCI.

Medical Treatment Failure for Symptomatic Vasospasm After Subarachnoid Hemorrhage Threatens Long-Term Outcome.

Background and Purpose- Symptomatic vasospasm is a common cause of morbidity and mortality after subarachnoid hemorrhage. We sought to identify predictors and the long-term impact of treatment failure with hypertensive therapy for symptomatic vasospasm. Methods- We performed a retrospective analysis of 1520 subarachnoid hemorrhage patients prospectively enrolled in the Columbia University SAH Outcomes Project between August 1996 and August 2012. One hundred ninety-eight symptomatic vasospasm patients were treated with vasopressors to raise arterial blood pressure, with and without volume expansion. Treatment response, defined as complete or near-complete resolution of the initial neurological deficit, was adjudicated in weekly meetings of the study team based on serial clinical examination after hypertensive treatment. Outcome was evaluated at 1 year with the modified Rankin Scale. Results- Twenty-one percent of the 198 patients who received hypertensive therapy did not respond to treatment. Treatment failure was associated with an increased risk of death or severe disability at 1 year (modified Rankin Scale score of 4-6; 62% versus 25%; P<0.001). Failure of medical therapy was also associated with an admission troponin I level >0.3 µg/L (64% versus 28%; P=0.001), aneurysm coiling (43% versus 20%; P=0.004), and involvement of >1 symptomatic vascular territory at onset (39% versus 22%; P=0.02). In multivariable analysis, treatment failure was independently associated only with troponin I elevation (adjusted odds ratio, 4.30; 95% CI, 1.69-11.09; P=0.002). Conclusions- Failure to respond to induced hypertension for symptomatic vasospasm threatens 1-year outcome. Subarachnoid hemorrhage patients with symptomatic vasospasm who have elevated initial troponin I levels, indicative of neurogenic cardiac injury, are at twice the risk of medical treatment failure. Expedited endovascular therapy should be considered in these patients.

Predicting delayed cerebral ischemia after subarachnoid hemorrhage using physiological time series data.

To develop and validate a prediction model for delayed cerebral ischemia (DCI) after subarachnoid hemorrhage (SAH) using a temporal unsupervised feature engineering approach, demonstrating improved precision over standard features. 488 consecutive SAH admissions from 2006 to 2014 to a tertiary care hospital were included. Models were trained on 80%, while 20% were set aside for validation testing. Baseline information and standard grading scales were evaluated: age, sex, Hunt Hess grade, modified Fisher Scale (mFS), and Glasgow Coma Scale (GCS). An unsupervised approach applying random kernels was used to extract features from physiological time series (systolic and diastolic blood pressure, heart rate, respiratory rate, and oxygen saturation). Classifiers (Partial Least Squares, linear and kernel Support Vector Machines) were trained on feature subsets of the derivation dataset. Models were applied to the validation dataset. The performances of the best classifiers on the validation dataset are reported by feature subset. Standard grading scale (mFS): AUC 0.58. Combined demographics and grading scales: AUC 0.60. Random kernel derived physiologic features: AUC 0.74. Combined baseline and physiologic features with redundant feature reduction: AUC 0.77. Current DCI prediction tools rely on admission imaging and are advantageously simple to employ. However, using an agnostic and computationally inexpensive learning approach for high-frequency physiologic time series data, we demonstrated that our models achieve higher classification accuracy.

Deriving the PRx and CPPopt from 0.2-Hz Data: Establishing Generalizability to Bedmaster Users.

OBJECTIVE: The objective was to explore the validity of industry-parameterized vital signs in the generation of pressure reactivity index (PRx) and optimal cerebral perfusion pressure (CPPopt) values. MATERIALS AND METHODS: Ten patients with intracranial pressure (ICP) monitors from 2008 to 2013 in a tertiary care hospital were included. Arterial blood pressure (ABP) and ICP were sampled at 240 Hz (of waveform data) and 0.2 Hz (of parameterized data produced by heuristic industry proprietary algorithms). 240-Hz ABP were filtered for pulse pressure and diastolic ABP within the limits of 20-150 mmHg. The PRx was calculated as Pearson's correlation coefficient using 10-s averages of ICP and ABP over a 5-min moving window with 80% overlap. For ease of comparison, we used the naming convention of BMx for PRx values derived from 0.2-Hz data. A 5-min median cerebral perfusion pressure (CPP) trend was calculated, PRx or BMx values divided and averaged into CPP bins spanning 5 mmHg. The minimum Y value (PRx or BMx) of the parabolic function fit to the resulting XY plot of 4 h of data was obtained, and updated every 1 min. Pearson's R correlations were calculated for each patient. Linear mixed-effects models were used with a random intercept to assess the overall correlation between the PRx (outcome) and the BMx (fixed effect) or the CPPopt-PRx (outcome) and the CPPopt-BMx (fixed effect). RESULTS: The overall correlation between the PRx and BMx was 0.78 based on the linear mixed effects models (p < 0.0001), and the overall correlation for the CPPopt-PRx and CPPopt-BMx based on the linear mixed effects models was 0.76 (p < 0.0001). One patient had low correlation of CPPopts derived from the PRx vs the BMx; this patient had the least number of hours of CPPopt data to compare. CONCLUSIONS: The BMx shows promise in CPPopt derivation against the validated PRx measure. If further developed, it could expand the capability of centers to derive CPPopt goals for use in clinical trials.