Why every ICU needs point-of-care EEG at the bedside

As a neurocritical care and stroke neurologist at a leading academic medical center, I’ve spent over two decades on the front lines, working with major community hospitals in triaging and receiving critically ill patients from the emergency department (ED) and intensive care unit (ICU). In my world, “Time is Brain” is not just a slogan but a mandate.

Whether it’s stroke or seizure or any acute brain injury, rapid detection of neurological distress is the difference between recovery and irreversible harm. In the ICU and ED, the beeping of bedside monitors guides the rhythm of care: heart rate, blood pressure, oxygen saturation. They are constant, visible, audible, and actionable.

And yet, for the organ that defines consciousness, identity and function (the brain), monitoring has historically been delayed, intermittent, or altogether absent. That paradigm is now changing in the technological age. We are entering a new era where rapid, accessible bedside brain monitoring is no longer science fiction, but real life. Electroencephalogram (EEG) is emerging as the next vital sign, transforming how we detect, triage, and treat neurological emergencies.

The hidden crisis of “silent” seizures

In the ICU and ED, altered mental status is a daily challenge, ranging from subtle confusion to profound coma. Beneath this clinical spectrum lies a frequently missed and dangerous reality. A considerable proportion of these patients suffer from nonconvulsive seizures (NCS) or nonconvulsive status epilepticus (NCSE). These are truly “silent” seizures where there are no convulsions, no warning signs, and no outward motor manifestations. They can only be detected through EEG.

The prevalence is alarming: Between 10 percent and 35 percent of patients who remain comatose after the return of spontaneous circulation (ROSC) following cardiac arrest experience seizures. In the context of sepsis, up to a third of patients with altered mentation may have seizures or periodic discharges. Without real-time EEG monitoring, these patients may be undiagnosed, untreated, and languish in the ICU. The consequences are profound: NCS and NCSE are associated with markedly worse outcomes, including a 14-fold increase in the odds of poor one-year survival and functional outcomes for those with post-anoxic status epilepticus, compared with those without.

Bridging the gap: from days to minutes

The traditional model of conventional EEG struggles to meet the needs of the acutely ill. Even in the best-resourced academic setting, conventional EEG remains a bottleneck. The process requires a specialized technician for setup, followed by delays while a sufficient data segment is recorded. Only then can the results be interpreted by neurologists with expertise in EEG and the results communicated to the clinical team. From EEG order to actionable insight, this process can take hours. In critically ill patients, this delay is unacceptable. The reality is that the majority of hospitals across the U.S. do not have timely access to conventional EEG or to specialists who can interpret the results. Even in an academic setting, getting a conventional EEG in the ED is always a challenge.

At our center, we’ve addressed this delay by utilizing point-of-care (POC) EEG that enables rapid bedside monitoring from the ED to the ICU. At our hospital system, we’ve integrated rapid EEG monitoring into our standard workflows: all EEGs at night and all ED EEGs now begin with a POC device. These can be set up within minutes, and the usage is aligned with national recommendations to initiate EEG monitoring within 60 minutes of suspected status epilepticus if there’s suspicion of ongoing seizures.

The POC EEG is paired with artificial intelligence (AI)-assisted interpretation, which converts complex EEG data into a simple quantitative measure of seizure burden that is actionable for bedside clinicians and nurses. More detailed analysis, including raw EEG to aid seizure localization and events to mark treatment, is accessible on a mobile device or computer. This transforms EEG into a true brain vital sign that’s immediately accessible and assists frontline teams with detecting pathology in real time and responding quickly.

Empowering the bedside team with real-time data

Now, automated bedside alerts and intuitive “seizure burden” displays provide the clinical team with immediate information for action. The seizure burden is expressed as a percentage, has a trend line, and tells the bedside team about how long the brain has been seizing in the preceding 5 minutes. It has a high negative predictive value for ruling out status epilepticus and seizures to aid patient triage. We can also see in real time how a patient responds to an anti-seizure medication, allowing us to titrate doses effectively without over-sedating the patient.

Clinical impact: cardiac arrest, stroke, and sepsis

The utility of this “brain vital sign” extends across various disease states. In cardiac arrest survivors, early EEG monitoring is essential because seizures can emerge quickly after resuscitation. American Heart Association (AHA) and other national guidelines now emphasize the prompt performance and interpretation of EEG for any patient not following commands post-ROSC.

In cases of ischemic stroke or subarachnoid hemorrhage, we use POC EEG when a patient’s mental status is depressed out of proportion to their imaging. Furthermore, in septic patients, certain EEG patterns, such as periodic discharges or absent reactivity, serve as powerful prognostic markers that help us predict in-hospital mortality.

A simpler workflow for health care providers

Integrating this technology has fundamentally improved our work-life balance and clinical efficiency. At our center, we’ve devised a simple color scheme for internal bedside nursing and physician workflow that guides the team about the urgency for treatment. For example, if the seizure burden is less than 10 percent, given the negative predictive value of 99 percent for ruling out definite and possible electrographic status epilepticus, we label this as “Green is clean.” If the seizure burden is greater than 90 percent, the specificity is 99 percent for detecting electrographic status and possible status epilepticus and this is labeled as “Red is med.” The recording that falls in the 10 percent to 90 percent zone is escalated for human review by the epilepsy team and is labeled as “Yellow is fellow.”

Approximately 80 percent of the EEGs that are ordered in our center’s ED and ICU turn out to be normal or slow activity (that is, green is clean), requiring only non-urgent review. By quickly identifying these “clean” cases at the bedside, we can avoid overmedicating these patients. We can also focus our intensive resources on the rest of the patients who experience concerning epileptiform patterns and status epilepticus and require immediate, life-saving care.

The future of the bedside

“AI Is Taking Away Jobs” is the headline almost every day. What we are seeing at my academic center is that it’s a supporting tool to help our epileptologists and our hospital staff ensure that patients get timely EEG when it matters most. It enables us to triage and redirect resources to the patients who need them. The goal of POC EEG is to ensure that no patient suffers from a treatable brain injury. By treating EEG as a vital sign available at the bedside 24/7 and interpretable in real time, we are finally giving the brain the same level of vigilance we give the heart. This paradigm shift saves more than just time; it saves lives and preserves the quality of life for our most vulnerable patients.

Chitra Venkatasubramanian is a neurologist specializing in neurocritical care and stroke. She is a clinical professor at Stanford University and practices at Stanford Health Care.

She spearheaded the implementation of rapid electroencephalography (EEG) at her institution and supports peer education at community and academic centers. She consults for Ceribell, Inc., and can be found on LinkedIn.