EEG for bedside monitoring: the intensivist's point of view.

EEG for bedside monitoring: the intensivist’s point of view - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Crit Care . 2026 Apr 17;30:195. doi: 10.1186/s13054-026-06023-8 Search in PMC Search in PubMed View in NLM Catalog Add to search EEG for bedside monitoring: the intensivist’s point of view Fabio Silvio Taccone Fabio Silvio Taccone 1 Department of Intensive Care, Hôpital Universitaire de Bruxelles (HUB), Université Libre de Bruxelles (ULB), Route de Lennik, 808, Brussels, 1070 Belgium Find articles by Fabio Silvio Taccone 1, ✉ , Taylan Ozkaya Taylan Ozkaya 1 Department of Intensive Care, Hôpital Universitaire de Bruxelles (HUB), Université Libre de Bruxelles (ULB), Route de Lennik, 808, Brussels, 1070 Belgium 2 Department of Anesthesia and Intensive Care unit, University of Trieste, Trieste, Italy Find articles by Taylan Ozkaya 1, 2 , Marta Baggiani Marta Baggiani 3 Anesthesiology and Intensive Therapy, San Gerardo Hospital, Monza, Italy Find articles by Marta Baggiani 3 , Frank A Rasulo Frank A Rasulo 4 Department of Anesthesiology and Intensive Care, University of Brescia, Brescia, Italy Find articles by Frank A Rasulo 4 Author information Article notes Copyright and License information 1 Department of Intensive Care, Hôpital Universitaire de Bruxelles (HUB), Université Libre de Bruxelles (ULB), Route de Lennik, 808, Brussels, 1070 Belgium 2 Department of Anesthesia and Intensive Care unit, University of Trieste, Trieste, Italy 3 Anesthesiology and Intensive Therapy, San Gerardo Hospital, Monza, Italy 4 Department of Anesthesiology and Intensive Care, University of Brescia, Brescia, Italy ✉ Corresponding author. Received 2026 Jan 24; Accepted 2026 Apr 8; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13097847  PMID: 41998712 Abstract Electroencephalography (EEG) is a powerful tool that can provide unique and real time insight into cerebral functioning in the context of acute brain injury in the intensive care unit (ICU), ranging from focal deficits to seizures and coma. Despite being a safe, relatively inexpensive, non-invasive and meaningful tool, EEG has not yet transitioned into a true bedside monitoring system in the ICU, as continuous EEG monitoring cannot realistically be provided to all ICU patients, and EEG implementation and interpretation remains heavily dependent on specialized personnel. In order to integrate EEG into routine ICU monitoring, two conditions must be fulfilled: first, the EEG montage should be adjusted to answer the specific clinical question; second, the presentation of EEG-derived information must be stratified and adapted to the healthcare professional interpreting it, from the inexperienced nurses and junior physicians to the experienced neurophysiologist. Integrating the EEG into the multimodal monitoring of critically ill patients would allow earlier detection of reversible brain insults, it would promote brain monitoring across different levels of expertise, and it could potentially expand EEG use with rapid data acquisition that could facilitate early identification and treatment of acute brain events, even outside the ICU. Keywords: Neuromonitoring, Non-invasive, EEG, Intensive Care, Brain injury Why EEG matters in the ICU Acute brain dysfunction is highly prevalent among critically ill patients, occurring either as a primary neurological insult or as a secondary consequence of systemic illness. Delirium, coma, seizures, hypoxic–ischemic brain injury, metabolic encephalopathy, and sedation-related cortical suppression are encountered daily in intensive care units (ICU) [ 1 , 2 ]. In this context, electroencephalography (EEG) represents one of the few tools capable of directly interrogating brain function in real time, offering insight that is otherwise inaccessible through clinical examination alone, particularly in sedated or unconscious patients [ 3 , 4 ]. Over the past two decades, the role of EEG, especially continuous EEG (cEEG), has expanded substantially. This evolution has been driven by the recognition that a large proportion of seizures in ICU patients are non-convulsive, lack clinical correlates, and remain undetected without electrophysiological monitoring [ 5 , 6 ]. Multiple high-impact studies have demonstrated that delayed identification of electrographic seizures or non-convulsive status epilepticus is associated with worse neurological outcomes [ 7 , 8 ]. Beyond seizure detection, EEG has emerged as a powerful diagnostic tool for the identification of delayed cerebral ischemia after subarachnoid hemorrhage, as a prognostic instrument in hypoxic-ischemic brain injury, sepsis-associated encephalopathy, traumatic brain injury and acute ischemic stroke and as a means of unmasking oversedation [ 9 – 14 ] (Fig. 1 ). Fig. 1. Open in a new tab Examples of relevant EEG patterns requiring rapid interpretation in critically ill patients. The upper panel (A) illustrates a 3-Hz spike-and-wave activity interspersed with rapid rhythms, maximal over the left frontocentral region, typical of ongoing seizures. Accurate identification of these patterns requires specific expertise, familiarity with standardized EEG terminology such as the American Clinical Neurophysiology Society (ACNS) nomenclature, and the ability to differentiate pathological activity from artifacts. The lower panel (B) demonstrates a burst–suppression pattern occurring after cardiac arrest, characterized by alternating periods of high-amplitude bursts and near-isoelectric suppression. A discontinuous EEG pattern characterized by background suppression and, in more advanced stages, burst–suppression may also occur during deep pharmacological sedation and should therefore be interpreted in the context of the patient’s clinical condition and exposure to sedative agents Accordingly, international guidelines increasingly recommended EEG monitoring for critically ill patients with unexplained altered consciousness, suspected seizures or severe acute brain injury [ 15 , 16 ]. From the intensivist’s standpoint, EEG is therefore conceptually appealing: it is non-invasive, physiologically meaningful, temporally rich, and potentially actionable. Yet, despite these advantages and strong recommendations, EEG has not fully transitioned into a true bedside monitoring modality in routine ICU practice. Is EEG really a bedside monitoring tool? In theory, EEG provides unmatched temporal resolution and continuous global assessment of cortical activity. In practice, however, its implementation in the ICU remains limited by logistical, technical and cultural barriers (Table 1 ), which prevent seamless integration into bedside decision-making. First, EEG implementation still depends largely on specialized personnel; indeed, conventional multichannel EEG requires trained technicians for the accurate electrode placement and maintenance. Although simplified montages and EEG caps have been developed [ 17 ], and in some centers ICU nurses are trained to apply them, this approach is not yet widespread. As a result, EEG initiation may be delayed by hours, precisely when early information would be most valuable. Second, the availability of EEG resources remains a major limiting factor to its extended use in ICU patients; cEEG monitoring cannot realistically be provided to all ICU patients, even in well-resourced centers [ 18 ], and intermittent EEG or simplified montages are therefore frequently adopted as pragmatic alternatives [ 19 ]. While existing literature suggested that cEEG was not superior to intermittent recordings in terms of global outcomes [ 19 , 20 ], the diagnostic and prognostic value of EEG montage is highly context-dependent, and the two approaches are not interchangeable. Third, EEG interpretation is not intuitive for most ICU clinicians. Unlike arterial pressure or oxygen saturation, EEG does not provide immediately understandable metrics. For many ICU teams, the EEG machine remains a “foreign body” at the bedside, which is present but disconnected from daily clinical reasoning. This lack of perceived ownership limits its potential influence on real-time management. While quantitative EEG (qEEG) tools and automated analyses are increasingly available [ 19 ], the reference standard remains visual inspection of raw EEG by experienced readers. This introduces an inherent delay between signal acquisition and clinical interpretation, particularly during night-shifts or weekends. Moreover, large multicenter analyses have shown only moderate inter-rater agreement among expert neurophysiologists in identifying ictal–interictal injury continuum patterns on EEG, mostly because of differences in diagnostic thresholds [ 21 ], which might further reduce the generalizability of EEG findings among different ICUs. Table 1. Common challenges for EEG monitoring in ICU patients Data collection Need for trained personnel for electrode placement and maintenance Technical challenges in non-cooperative or delirious patients Risk of skin irritation with prolonged recordings Scalp injuries or surgical dressings Quality of the EEG signal Use of sedatives ICU environment with multiple sources of electrical and mechanical interference Data reporting Quantitative metrics for interpretation Not real continuous reporting of EEG findings Combination with video recording (to interpret abnormal movements) Integration Limited options to combine EEG data with other neuromonitoring tools Interpretation Neurophysiologist is required to review EEG reading Fair agreement on the interpretation of some EEG findings Open in a new tab Thus, despite its physiological richness, EEG, in its current form, exerts an intermittent rather than continuous clinical impact, even when cEEG is technically running, and does not fulfill the essential criteria of a bedside monitoring tool: immediate availability, continuous interpretability and direct actionability for the treating team. The intensivist’s question: what do we really need from EEG? A key barrier to EEG integration is the implicit assumption that meaningful EEG use requires always and only neurophysiologist-level expertise. From the intensivist’s perspective, this assumption is hardly applicable. The history of critical care monitoring offers a useful parallel. ICU physicians routinely acquire and interpret bedside echocardiography without being cardiologists. The objective is not to identify rare cardiomyopathies or subtle valvular abnormalities, but rather to address focused, clinically relevant questions, such as the presence of low cardiac output, pericardial tamponade, or acute right ventricular failure, that directly inform immediate management [ 22 ]. In a similar manner, lung ultrasonography is now routinely performed by intensivists without the need for radiologist-level expertise [ 22 ], and bronchoscopic procedures, including bronchoalveolar lavage [ 23 ], are commonly and safely carried out in the ICU without direct involvement of a pulmonologist. EEG could follow the same trajectory. The objective is not to transform intensivists into neurophysiologists, which is an unrealistic goal, given that EEG expertise requires the interpretation of thousands of EEG recordings, which would be incompatible with routine ICU practice, but rather to equip them with the ability of recognizing a limited set of high-value EEG patterns that carry immediate and actionable clinical significance. From an intensivist’s standpoint, four EEG features are particularly relevant and sufficient for most bedside decisions [ 24 ]: a) background activity (i.e., continuous vs. discontinuous vs. suppressed); asymmetry (suggesting focal structural or ischemic injury); burst suppression (spontaneous or drug-induced); ictal or highly epileptiform activity. These elements capture the majority of urgent, actionable EEG information in the ICU. They are conceptually simple, clinically meaningful, and already strongly linked to outcomes in multiple high-impact studies. Several studies have demonstrated that even a brief, structured training program can substantially improve the ability of ICU physicians and nurses to correctly classify key EEG features [ 25 , 26 ], including background activity, seizure patterns, hemispheric asymmetry, and common artifacts. Making EEG a true bedside monitoring tool: a tiered information model To transform EEG into a genuine bedside monitor, two fundamental conditions must be met: first, the EEG montage should be tailored to the specific clinical question; second, the presentation of EEG-derived information must be adapted to the user’s level of expertise. EEG montage: a question-driven approach The diagnostic and prognostic value of EEG montage is highly context-dependent. For instance, when the primary objective is seizure detection or monitoring for DCI, cEEG is inherently superior [ 10 ], as the onset of these events is unpredictable and intermittent recordings may simply miss clinically relevant episodes. In contrast, for outcome prediction, such as neurological prognostication after cardiac arrest, short, high-quality intermittent EEG recordings may provide information comparable to continuous monitoring, particularly when assessing highly malignant patterns [ 27 ]. A similar principle applies to electrode montage selection. Standard full 10/20 electrode placement is essential when spatial resolution is critical, such as for seizure localization or detection of focal ischemic changes; however, simplified montages may be sufficient for other purposes, including identification of burst suppression related to excessive sedation, or highly malignant EEG patterns following hypoxic–ischemic brain injury [ 28 ]. These observations underscore that EEG monitoring should not be conceptualized as a single, uniform tool, but rather as a flexible diagnostic strategy tailored to the clinical question. Accordingly, centers with EEG capability should aim to maintain access to a spectrum of EEG modalities (i.e., continuous and intermittent, full and simplified montages), allowing clinicians to select the most appropriate approach based on the patient’s condition and the intended clinical objective, rather than defaulting to a one-size-fits-all solution. Where possible, each ICU bed could be equipped with qEEG capabilities for sedation monitoring, whenever indicated, while full-montage EEG systems would be reserved for patients requiring cEEG for specific diagnostic or monitoring indications. This approach is likely to substantially expand access to EEG monitoring to a greater proportion of critically ill patients. EEG output: hierarchical and customizable Rather than relying on a single, highly complex display designed exclusively for specialists, EEG output in the ICU should be stratified and adapted to the level of expertise of the healthcare professional interpreting it. The first level of interpretation could consist of a binary safety alert system, intended for EEG-inexperienced nurses and first-line responders, such as junior physicians on duty (Fig. 2 ). At the most basic level, EEG information could be simplified into a “red alert” signal, displayed directly on the patient’s standard monitoring screen, in a manner analogous to conventional vital sign alarms, such as oxygen saturation or arterial blood pressure. These alarms would not display waveforms, but rather indicate danger states using intuitive visual cues (e.g., a brain icon turning red, indicating an abnormality in the right hemisphere, left hemisphere, or both). This level would require no specific EEG expertise and would enable bedside nurses to be immediately alerted to potentially significant cerebral abnormalities, prompting timely notification of the treating physicians, in the same way alarms currently signal critical deviations in blood pressure or oxygenation. Fig. 2. Open in a new tab Conceptual framework illustrating the progressive levels of EEG interpretation across acute care settings. In the intensive care unit, EEG information is delivered according to the user’s level of expertise. At the most basic level, untrained nursing staff are provided with a simplified safety alarm integrated into the standard patient monitor, displayed as a brain icon indicating the lateralization of a detected abnormality. At the next level, non-neurologist physicians can access additional information specifying the nature of the abnormal pattern (e.g., suspected ictal activity in one hemisphere). EEG-trained ICU nurses and physicians may then review quantitative EEG (qEEG) outputs, including background trends, hemispheric asymmetry indices, burst suppression ratio, and seizure probability metrics. These successive layers of interpretation are supported by artificial intelligence–based signal analysis and prioritization. At the highest level, neurophysiologists perform comprehensive review of qEEG trends and raw EEG signals to provide definitive interpretation and formal reporting. While interpretative complexity and accuracy increase across levels, the initial alarm layer is designed to operate continuously. The EEG montage should be tailored to the clinical question and the anticipated pathological pattern The second level would add minimal contextual information, such as the type of abnormality that triggers the alarm. This could be, as an example, repeated ictal activity on the right hemisphere, or an increased rate of burst suppression, as due to excessive sedation. This level of information would be accessible to non-neurologist physicians without formal training in EEG interpretation and would trigger two possible actions: either contacting an ICU team member with basic EEG training (e.g., recognition of background patterns, asymmetry, burst suppression, ictal or highly epileptiform activity, and common artifacts) or directly alerting the on-call neurophysiologist for prompt EEG review. Importantly, no urgent clinical decisions, such as neuroimaging, initiation of antiseizure therapy, or escalation of other treatments, would be made by untrained healthcare providers based solely on this alarm. Instead, all subsequent clinical actions would be deferred to personnel with adequate expertise to interpret the underlying EEG data supporting the alert. However, a critical challenge remains: ensuring that the alarm signal is both reliable and clinically meaningful. An effective system must achieve high sensitivity to detect clinically relevant EEG abnormalities while maintaining sufficient specificity to avoid excessive false-positive alerts. An alarm that fails to identify significant pathological patterns undermines patient safety, whereas one that triggers too frequently risks alarm fatigue, desensitization of staff, and loss of clinical credibility. Striking this balance is essential for EEG-based alerts to be safely and effectively integrated into routine bedside monitoring in the ICU. Recent advances in artificial intelligence (AI) have made this approach technically feasible. Several high-quality studies have demonstrated that machine-learning algorithms can detect seizures and other critical EEG patterns with performance comparable to expert neurophysiologists [ 29 , 30 ]. Importantly, AI does not need to replace human expertise; its primary value lies in continuous surveillance, fatigue-free pattern recognition, and real-time alerting. AI-driven EEG systems should therefore be integrated into modern neurophysiological monitoring to enable automated detection of clinically relevant EEG patterns, while simultaneously providing real-time assessment of signal quality. Indicators such as excessive artifacts, loss of electrode contact, or increased electrode impedance could prompt targeted manual intervention, including electrode repositioning or montage verification, thereby ensuring data reliability and preserving the clinical usefulness of EEG as a bedside monitoring tool. The third level of interpretation relies on qEEG trend analyses and is intended for ICU clinicians with dedicated EEG training. These representations allow clinicians to follow cerebral dynamics continuously without requiring full raw waveform interpretation. At this level, more advanced displays can be provided, including compressed spectral arrays (spectrograms), EEG amplitude trends or alpha/delta ratio with automatic analysis of symmetry, burst suppression ratio and ictal activities. Spectrograms offer a compact visualization of frequency distribution over time and are particularly useful for detecting progressive background slowing (i.e., excessive sedation or increased intracranial hypertension) or burst suppression events (i.e., black bars) [ 31 ]. Spectrograms are frequently complemented by quantitative descriptors of cortical activity, such as the Spectral Edge Frequency (SEF), which represents the frequency below which a predefined proportion (commonly 90% or 95%) of the total EEG power is contained, and of artefacts due to muscular contraction, such as the electromyographic (EMG) activity. Thus, lower SEF values reflect a shift toward slower frequencies and are therefore indicative of deeper levels of cerebral suppression, while apparent increases in high EEG frequency or power on the spectrogram may primarily reflect muscular contamination, particularly in the presence of elevated EMG values. EEG amplitude trends can help identify global suppression, sudden attenuation suggestive of ischemia, or recovery of cortical activity after sedation reduction or resuscitation. In this context, the alpha/delta ratio can serve as a surrogate of cortical integrity and functional recovery, with reductions often indicating worsening encephalopathy or focal ischemia [ 32 ]. Burst suppression ratio provides an objective measure of deep cerebral suppression, enabling titration of sedative or anesthetic agents and early recognition of excessive suppression [ 33 ]. Longitudinal trend graphs facilitate the identification of gradual changes that may be clinically silent but pathologically relevant. Ictal activity is recognized through characteristic dynamic changes in frequency, amplitude, and rhythmicity that evolve over time. Typical features include a sudden increase in power within specific frequency bands (often beta, alpha, or theta), the emergence of rhythmic, repetitive patterns, and a progressive evolution in frequency, amplitude, or spatial distribution, hallmarks that distinguish seizures from static background abnormalities [ 34 ]. On spectrograms, seizures commonly appear as focal or hemispheric “hot spots” of increased power that may spread or shift in frequency, while amplitude trend displays show abrupt and sustained increases in signal magnitude. Importantly, the temporal evolution of these changes is critical, as ictal patterns typically demonstrate gradual buildup and resolution, in contrast to artifacts, which tend to be abrupt, transient, and non-evolving. At this level, clinicians are expected to recognize common artifacts (e.g., movement, muscle activity, electrical interference) and to interpret trends within the broader clinical context, rather than to perform detailed waveform analysis [ 35 ]. This approach mirrors the use of advanced hemodynamic monitoring, which is selectively applied by trained clinicians to guide individualized management rather than to replace expert interpretation. Finally, access to the full raw EEG signal, including adjustable montages, filters, and time scales, remains indispensable for definitive interpretation, diagnosis, and formal reporting. This highest level of analysis should remain within the expertise of neurophysiologists, who can be promptly alerted through upstream alarm systems when urgent review is required. Beyond real-time interpretation, the neurophysiology team plays a central role in sustaining the clinical value of EEG in the ICU by organizing structured educational programs for ICU physicians and nurses, reviewing previous recordings, facilitating multidisciplinary case discussions, and ensuring appropriate reassessment of diagnostic and therapeutic decisions informed by earlier levels of EEG alarm generation and interpretation. As expert EEG interpretation retains a certain degree of subjectivity [ 21 ], in the future, advances in AI–assisted analysis may help standardize interpretation and potentially complement or partially replace human expertise. Moreover, the involvement of a trained neurophysiologist may facilitate the assessment of EEG changes in response to external stimuli, commonly referred to as “EEG reactivity”, which represents an important area of investigation in critically ill patients and provides a dynamic interpretation of EEG activity, reflecting the functional integrity of cortical networks and their connectivity. Why this matters ? Transforming EEG into a true bedside monitor would have several major clinical benefits. First, it would enable earlier detection of reversible brain insults, such as non-convulsive seizures, metabolic encephalopathy, or excessive sedation. Time is brain, and delays in recognition often translate into worse outcomes [ 36 , 37 ]. Second, it would promote shared ownership of brain monitoring. When nurses and intensivists actively engage with EEG data, the technology becomes integrated into daily care rather than remaining peripheral. Third, clinicians would become more familiar with the cerebral “signatures” of commonly used sedatives, which differ substantially; while propofol is typically associated with increasing slow delta and alpha oscillations on the spectrogram as sedation deepens and with burst suppression at excessive doses, dexmedetomidine produces a sleep-like pattern characterized by slow-delta activity, ketamine generates a more activated tracing, with prominent high-beta and low-gamma activity, and volatile agents share propofol-like slow-alpha features at lower concentrations, but higher doses add theta activity and broaden the spectrographic power distribution [ 38 ]. Forth, as EEG technology evolves toward more automated interpretation, improved signal processing and AI-assisted analysis may enable wider assessment of brain function in unresponsive patients, supporting earlier recognition of consciousness or potentially help to better assess neurological prognosis [ 39 , 40 ]. Fifth, EEG would be definitely integrated into a multimodal neuromonitoring strategy for patients with acute brain injury, as combining electrophysiological signals with clinical, imaging, and other vital signs (i.e. intracranial pressure or brain oxygenation) can improve diagnostic accuracy and prognostic performance in critically ill patients [ 41 ]. Finally, it could potentially favor the expansion of EEG use to patients with acute brain dysfunction in an earlier stage and even beyond the ICU. In the emergency department, rapid EEG acquisition using a similar approach than in the ICU could favor an early EEG implementation and facilitate rapid identification of seizures, as well as detection of focal abnormalities suggestive of acute ischemic stroke or intracranial hemorrhage in patients presenting with sudden neurological deterioration. The concept of point-of-care EEG (POC-EEG) in the ED parallels the evolution observed with bedside ultrasound and focused echocardiography: a rapid, targeted diagnostic tool designed to answer time-critical clinical questions rather than to provide exhaustive neurophysiological assessment [ 42 ]. Beyond diagnosis, POC-EEG may also contribute to early prognostic stratification and treatment planning. For example, in post-cardiac arrest patients arriving at the ED, early EEG assessment could help identify patients with preserved cortical activity who may benefit from early controlled awakening, while avoiding unnecessary sedation in those with malignant patterns [ 43 , 44 ]. As technology advances, integration of POC-EEG with tele-neurophysiology platforms may further enhance its utility, enabling real-time expert support across institutions with variable access to neurophysiology services. Looking ahead, simplified EEG systems integrated with telemedicine platforms may extend this capability to the prehospital setting, allowing transmission of basic cerebral information to receiving hospitals. Such data could inform destination decisions (e.g., focal abnormalities prompting direct transfer to centers with neuroradiology and neurosurgical capabilities) or enable pre-arrival therapeutic interventions, such as early antiseizure treatment to prevent progression to refractory status epilepticus. Potential challenges Importantly, several potential challenges to this approach need to be mentioned. First, future multicenter studies will be necessary to evaluate the generalizability of real-time EEG-based monitoring approaches across different acquisition systems and montages, and the impact of such monitoring on clinical practices and clinically relevant outcomes. Nevertheless, the implementation of EEG as a bedside monitoring tool remains essential, as monitoring strategies are most effective when clinicians clearly understand the physiological meaning of the signals and can therefore incorporate them into monitoring-guided clinical decision-making and patient management. Second, although expanding EEG monitoring across ICU beds and using multiple montages could enhance neuromonitoring capabilities, widespread implementation remains challenging due to the non-negligible cost of EEG systems and infrastructure. Moreover, the risk of alarm fatigue, particularly in environments already characterized by extensive multimodal monitoring, should be considered. Strategies aimed at minimizing false-positive alerts, such as improved artifact detection, signal quality optimization, and careful customization of alarm thresholds, will therefore be essential to ensure that EEG-based alerts remain clinically meaningful and actionable [ 45 , 46 ]. Third, the training required for ICU staff to accurately interpret increasingly complex EEG data represents an additional barrier, underscoring the need for pragmatic and scalable approaches to EEG integration in critical care and the evaluation of their potential implementation in clinical practice. Also, although qEEG trends represent a valuable bedside monitoring tool, particularly for nursing staff in the early detection of seizures, vasospasm, or delayed cerebral ischemia, their effective implementation in routine ICU practice requires structured educational programs. Dedicated training initiatives for bedside nurses and other healthcare professionals are essential to support reliable pattern recognition, appropriate interpretation, and timely escalation within established clinical workflows. Forth, although EEG reactivity to stimuli, such as auditory or somatosensory inputs, has shown potential diagnostic and prognostic value, its clinical application remains limited by variable reproducibility across studies and centers [ 47 ]. In addition, the absence of standardized stimulation protocols further complicates its integration into routine bedside monitoring [ 48 , 49 ]. Future research should therefore aim to develop standardized methods for the evaluation and quantification of EEG reactivity, which could ultimately be incorporated into the third level of EEG interpretation following appropriate training and the identification of reliable qEEG metrics. Finally, future AI-driven EEG monitoring systems in the ICU should move beyond simple alarm-based alerts and evolve toward platforms that provide transparent, clinically interpretable outputs. In addition to detecting abnormal patterns, these systems should explicitly identify the EEG features underpinning algorithmic decisions, such as rhythmic and periodic discharges, seizure probability trajectories, background organization, or reactivity patterns, thereby facilitating clinician understanding and fostering trust in automated outputs. The importance of interpretability is increasingly recognized, as “black-box” models may limit adoption in high-stakes environments such as neurocritical care, where clinical decisions require both accuracy and explainability. Recent advances in machine learning and deep learning have demonstrated the feasibility of automated EEG interpretation. Large-scale models trained on tens of thousands of recordings have achieved diagnostic performance comparable to expert readers in classifying EEG abnormalities [ 30 ]. In the ICU setting, convolutional neural networks and other classifier-based approaches have shown promising results for early detection of seizures and other critical EEG patterns, including non-convulsive events that are frequently missed without continuous monitoring [ 50 ]. In patients at risk of DCI after SAH, machine learning–based approaches could enable rapid, automated, and multidimensional analysis of EEG data by integrating a wide range of features, including spectral characteristics, temporal dynamics, and spatial patterns, which resulted in a high predictive value for DCI [ 51 ]. However, these technologies are not yet widely available, and their implementation is currently limited by a lack of standardization, as well as practical barriers including cost, infrastructure requirements, validation across diverse patient populations, and integration into existing clinical workflows. Conclusions From the intensivist’s perspective, EEG holds immense untapped potential. It already provides unique insight into brain function, yet remains underutilized as a real-time bedside tool due to logistical, interpretative and cultural barriers. The solution is not to demand that every intensivist become a neurophysiologist, but to redesign EEG systems around clinical usability: minimal essential information, intelligent alarms, AI-assisted interpretation, and graduated access to complexity. If EEG can follow the path of echocardiography, moving from a specialist-only test to a minimal core ICU competency, it may finally fulfill its promise as a central pillar of multimodal monitoring in critically ill patients. Acknowledgements None. Abbreviations DCI Delayed cerebral ischemia EEG Electroencephalography cEEG Continuous electroencephalography qEEG Quantitative electroencephalography SEF Spectral edge frequency Author contributions FST and TO drafted the first version of the manuscript. FAR and MB critically revised the final version of the manuscript. Funding None received. Data availability No datasets were generated or analysed during the current study. Declarations Ethical approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests FST is Scientific Advisor for Neuroptics and Nihon Khoden. FAR received lecture fees from MASIMO. Other authors have no conflict of interest to declare. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References 1. Stollings JL, Kotfis K, Chanques G, Pun BT, Pandharipande PP, Ely EW. Delirium in critical illness: clinical manifestations, outcomes, and management. Intensive Care Med. 2021;47(10):1089–103. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Jeong H, Park SH, Choo YH, Kang DW, Kim YS, Yang BSK, et al. Epidemiological Patterns and Variability in Acute Brain Injury: A Multicenter Registry Analysis in South Korea’s Neurocritical Care Units. Neurocrit Care. 2025;43(3):810–23. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Alkhachroum A, Appavu B, Egawa S, Foreman B, Gaspard N, Gilmore EJ, et al. Electroencephalogram in the intensive care unit: a focused look at acute brain injury. Intensive Care Med. 2022;48(10):1443–62. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Rubinos C, Bruzzone MJ, Viswanathan V, Figueredo L, Maciel CB, LaRoche S. Electroencephalography as a Biomarker of Prognosis in Acute Brain Injury. Semin Neurol. 2023;43(5):675–88. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Claassen J, Mayer SA, Kowalski RG, Emerson RG, Hirsch LJ. Detection of electrographic seizures with continuous EEG monitoring in critically ill patients. Neurology. 2004;62(10):1743–8. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Karki B, Shrestha PS, Shrestha N, Shilpakar O, Acharya SP, Neupane A. Prevalence of non-convulsive seizures and electroencephalographic abnormalities in critically ill patients—A retrospective observational study. Epilepsia Open. 2024;9(1):325–32. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Vespa PM, McArthur DL, Xu Y, Eliseo M, Etchepare M, Dinov I, et al. Nonconvulsive seizures after traumatic brain injury are associated with hippocampal atrophy. Neurology. 2010;75(9):792–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Vespa PM, O’Phelan K, Shah M, Mirabelli J, Starkman S, Kidwell C, et al. Acute seizures after intracerebral hemorrhage. Neurology. 2003;60(9):1441–6. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Rosenthal ES, Biswal S, Zafar SF, O’Connor KL, Bechek S, Shenoy AV, et al. Continuous electroencephalography predicts delayed cerebral ischemia after subarachnoid hemorrhage: A prospective study of diagnostic accuracy. Ann Neurol. 2018;83(5):958–69. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Guo Y, Fang S, Wang J, Wang C, Zhao J, Gai Y. Continuous EEG detection of DCI and seizures following aSAH: a systematic review. Br J Neurosurg. 2020;34(5):543–8. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Westhall E, Rossetti AO, van Rootselaar AF, Wesenberg Kjaer T, Horn J, Ullén S, et al. Standardized EEG interpretation accurately predicts prognosis after cardiac arrest. Neurology. 2016;86(16):1482–90. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Pantzaris ND, Platanaki C, Tsiotsios K, Koniari I, Velissaris D. The Use of Electroencephalography in Patients with Sepsis: a Review of The Literature. J Transl Int Med. 2021;9(1):12–6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Benghanem S, Novy J, Schindler KA, Rüegg S, Alvarez V, Rossetti AO. Unreactive EEG Background Is Associated With Unfavorable Outcome in Patients With Disorders of Consciousness of Various Etiologies: An Adult Cohort Study. Eur J Neurol. 2025;32(8):e70312. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Benghanem S, Kubis N, Gayat E, Loiodice A, Pruvost-Robieux E, Sharshar T, et al. Prognostic value of early EEG abnormalities in severe stroke patients requiring mechanical ventilation: a pre-planned analysis of the SPICE prospective multicenter study. Crit Care. 2024;28(1):173. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Claassen J, Taccone FS, Horn P, Holtkamp M, Stocchetti N, Oddo M. Recommendations on the use of EEG monitoring in critically ill patients: consensus statement from the neurointensive care section of the ESICM. Intensive Care Med. 2013;39(8):1337–51. [ DOI ] [ PubMed ] [ Google Scholar ] 16. Brophy GM, Bell R, Claassen J, Alldredge B, Bleck TP, Glauser T, et al. Guidelines for the Evaluation and Management of Status Epilepticus. Neurocrit Care. 2012;17(1):3–23. [ DOI ] [ PubMed ] [ Google Scholar ] 17. Ibrahim ZH, Chari G, Abdel Baki S, Bronshtein V, Kim MR, Weedon J, et al. Wireless multichannel electroencephalography in the newborn. J Neonatal Perinat Med. 2016;9(4):341–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Park A, Boyd JG. EEG utilization in the medical/surgical ICU: a single centre prospective observational study. Intensive Care Med. 2015;41(10):1869–70. [ DOI ] [ PubMed ] [ Google Scholar ] 19. Fenter H, Rossetti AO, Beuchat I. Continuous versus Routine Electroencephalography in the Intensive Care Unit: A Review of Current Evidence. Eur Neurol. 2024;87(1):17–25. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Rossetti AO, Schindler K, Sutter R, Rüegg S, Zubler F, Novy J, et al. Continuous vs Routine Electroencephalogram in Critically Ill Adults With Altered Consciousness and No Recent Seizure: A Multicenter Randomized Clinical Trial. JAMA Neurol. 2020;77(10):1225–32. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Jing J, Ge W, Struck AF, Fernandes MB, Hong S, An S, et al. Interrater Reliability of Expert Electroencephalographers Identifying Seizures and Rhythmic and Periodic Patterns in EEGs. Neurology. 2023;100(17):e1737–49. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Robba C, Wong A, Poole D, Al Tayar A, Arntfield RT, Chew MS, et al. Basic ultrasound head-to-toe skills for intensivists in the general and neuro intensive care unit population: consensus and expert recommendations of the European Society of Intensive Care Medicine. Intensive Care Med. 2021;47(12):1347–67. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Solidoro P, Corbetta L, Patrucco F, Sorbello M, Piccioni F, D’amato L, et al. Competences in bronchoscopy for Intensive Care Unit, anesthesiology, thoracic surgery and lung transplantation. Panminerva Med. 2019;61(3):367–85. [ DOI ] [ PubMed ] [ Google Scholar ] 24. Hirsch LJ, Fong MWK, Leitinger M, LaRoche SM, Beniczky S, Abend NS, et al. American Clinical Neurophysiology Society’s Standardized Critical Care EEG Terminology: 2021 Version. J Clin Neurophysiol. 2021;38(1):1–29. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Citerio G, Patruno A, Beretta S, Longhi L, Frigeni B, Lorini L. Continuous Quantified EEG in NeuroIntensive Care (CrazyEEG) Group. Implementation of continuous qEEG in two neurointensive care units by intensivists: a feasibility study. Intensive Care Med. 2017;43(7):1067–8. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Taran S, Ahmed W, Pinto R, Bui E, Prisco L, Hahn CD, et al. Educational initiatives for electroencephalography in the critical care setting: a systematic review and meta-analysis. Can J Anaesth. 2021;68(8):1214–30. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Fatuzzo D, Beuchat I, Alvarez V, Novy J, Oddo M, Rossetti AO. Does continuous EEG influence prognosis in patients after cardiac arrest? Resuscitation. 2018;132:29–32. [ DOI ] [ PubMed ] [ Google Scholar ] 28. Lybeck A, Cronberg T, Borgquist O, Düring JP, Mattiasson G, Piros D, et al. Bedside interpretation of simplified continuous EEG after cardiac arrest. Acta Anaesthesiol Scand. 2020;64(1):85–92. [ DOI ] [ PubMed ] [ Google Scholar ] 29. Jing J, Sun H, Kim JA, Herlopian A, Karakis I, Ng M, et al. Development of Expert-Level Automated Detection of Epileptiform Discharges During Electroencephalogram Interpretation. JAMA Neurol. 2020;77(1):103–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Tveit J, Aurlien H, Plis S, Calhoun VD, Tatum WO, Schomer DL, et al. Automated Interpretation of Clinical Electroencephalograms Using Artificial Intelligence. JAMA Neurol. 2023;80(8):805–12. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Ng MC, Jing J, Westover MB. A Primer on EEG Spectrograms. J Clin Neurophysiol. 2022;39(3):177–83. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Foreman B, Claassen J. Quantitative EEG for the detection of brain ischemia. Crit Care. 2012;16(2):216. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Andresen JM, Girard TD, Pandharipande PP, Davidson MA, Ely EW, Watson PL. Burst suppression on processed electroencephalography as a predictor of postcoma delirium in mechanically ventilated ICU patients. Crit Care Med. 2014;42(10):2244–51. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 34. Sharma S, Nunes M, Alkhachroum A. Adult Critical Care Electroencephalography Monitoring for Seizures: A Narrative Review. Front Neurol. 2022;13:951286. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. Amin U, Nascimento FA, Karakis I, Schomer D, Benbadis SR. Normal variants and artifacts: Importance in EEG interpretation. Epileptic Disord. 2023;25(5):591–648. [ DOI ] [ PubMed ] [ Google Scholar ] 36. Gururangan K, Kozak R, Dorriz PJ. Time is brain: detection of nonconvulsive seizures and status epilepticus during acute stroke evaluation using point-of-care electroencephalography. J Stroke Cerebrovas Dis. 2025;34(1):108116. [ DOI ] [ PubMed ] [ Google Scholar ] 37. Vespa PM, Olson DM, John S, Hobbs KS, Gururangan K, Nie K, et al. Evaluating the Clinical Impact of Rapid Response Electroencephalography: The DECIDE Multicenter Prospective Observational Clinical Study. Crit Care Med. 2020;48(9):1249–57. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Purdon PL, Sampson A, Pavone KJ, Brown EN. Clinical Electroencephalography for Anesthesiologists: Part I: Background and Basic Signatures. Anesthesiology. 2015;123(4):937–60. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Manasova D, Belloli LML, Rosenfelder MJ, Willacker L, Fló Rama E, Valota C et al. Multimodal multicentre investigation of diagnostic and prognostic markers in disorders of consciousness. Brain 2026, Epub ahead of print. 10.1093/brain/awaf412 [ DOI ] [ PMC free article ] [ PubMed ] 40. Pham SDT, Keijzer HM, Ruijter BJ, Seeber AA, Scholten E, Drost G, et al. Outcome Prediction of Postanoxic Coma: A Comparison of Automated Electroencephalography Analysis Methods. Neurocrit Care. 2022;37(Suppl 2):248–58. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Rohaut B, Calligaris C, Hermann B, Perez P, Faugeras F, Raimondo F, et al. Multimodal assessment improves neuroprognosis performance in clinically unresponsive critical-care patients with brain injury. Nat Med. 2024;30(8):2349–55. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Kozak R, Gururangan K, Dorriz PJ, Kaplan M. Point-of-care electroencephalography enables rapid evaluation and management of non-convulsive seizures and status epilepticus in the emergency department. J Am Coll Emerg Physicians Open. 2023;4(4):e13004. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Tjepkema-Cloostermans MC, da Silva Lourenço C, Ruijter BJ, Tromp SC, Drost G, Kornips FHM, et al. Outcome Prediction in Postanoxic Coma With Deep Learning. Crit Care Med. 2019;47(10):1424–32. [ DOI ] [ PubMed ] [ Google Scholar ] 44. Pelentritou A, Nguissi NAN, Iten M, Haenggi M, Zubler F, Rossetti AO, et al. The effect of sedation and time after cardiac arrest on coma outcome prognostication based on EEG power spectra. Brain Commun. 2023;5(4):fcad190. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Shaoru C, Hui Z, Su W, Ruxin J, Huiyi Z, Hongmei Z, et al. Determinants of Medical Equipment Alarm Fatigue in Practicing Nurses: A Systematic Review. SAGE Open Nurs. 2023;9:23779608231207227. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Kaleem S, Kang JH, Sahgal A, Hernandez CE, Sinha SR, Swisher CB. Electrographic Seizure Detection by Neuroscience Intensive Care Unit Nurses via Bedside Real-Time Quantitative EEG. Neurol Clin Pract. 2021;11(5):420–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Admiraal MM, van Rootselaar AF, Horn J. Electroencephalographic reactivity testing in unconscious patients: a systematic review of methods and definitions. Eur J Neurol. 2017;24(2):245–54. [ DOI ] [ PubMed ] [ Google Scholar ] 48. Caroyer S, Depondt C, Rikir E, Mavroudakis N, Peluso L, Taccone FS, et al. Assessment of a standardized EEG reactivity protocol after cardiac arrest. Clin Neurophysiol. 2021;132(7):1687–93. [ DOI ] [ PubMed ] [ Google Scholar ] 49. Tsetsou S, Novy J, Oddo M, Rossetti AO. EEG reactivity to pain in comatose patients: Importance of the stimulus type. Resuscitation. 2015;97:34–7. [ DOI ] [ PubMed ] [ Google Scholar ] 50. Degano G, Quintard H, Kleinschmidt A, Francini N, Sarbu OE, De Stefano P. ICU-EEG Pattern Detection by a Convolutional Neural Network. Ann Clin Transl Neurol. 2025;12(11):2295–304. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Zheng WL, Kim JA, Elmer J, Zafar SF, Ghanta M, Moura Junior V, Patel A, Rosenthal E, Brandon Westover M. Automated EEG-based prediction of delayed cerebral ischemia after subarachnoid hemorrhage. Clin Neurophysiol. 2022;143:97–106. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Data Availability Statement No datasets were generated or analysed during the current study. Articles from Critical Care are provided here courtesy of BMC ACTIONS View on publisher site PDF (1.4 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top