After seeing its impact in more than 1,000 patients, I can confidently say: AI-powered point-of-care EEG is transforming how we deliver neurological care. By removing barriers to access, it’s helping democratize seizure diagnosis across every part of the hospital.
Fast access to EEG for emergency and critical care patients has long been limited by logistical challenges. While we’ve made huge strides in rapid stroke response and post-cardiac arrest care, patients with neurological emergencies are still too often left waiting – sometimes for hours or even days for EEG.1234 The result? Delayed diagnoses, avoidable complications, and missed opportunities to improve outcomes.
As Chair of Neurology at a major academic medical center, I’ve seen how limited EEG access can affect care systemwide. Machines weren’t always available. Technologists couldn’t always respond quickly, especially at night. These delays affected patients far beyond the epilepsy unit and were even more pronounced at our partner sites, where neurology support is limited.
This matters deeply for patients suspected of having non-convulsive seizures. These events don’t involve visible convulsions. Instead, they may present subtly as confusion, staring, or agitation – symptoms that can easily be mistaken for sedation, encephalopathy, or post-ictal state. Without treatment, non-convulsive seizures can evolve into status epilepticus and cause irreversible brain injury. All-cause mortality rates for non-convulsive status epilepticus have been reported as high as 30%.5
Like many hospital systems, we needed a way to bring timely EEG to the bedside and make it usable by any trained provider. That’s why we implemented point-of-care EEG equipped with AI.
A Critical Need Across Departments
Many critically ill patients present with altered mental status and other risk factors for non-convulsive seizures. Studies suggest that up to 40% of ICU patients with altered consciousness are experiencing non-convulsive seizures.6 7 Yet prior to adopting point-of-care EEG, we had no real-time way to rule them in or out.
When we launched this point-of-care EEG system, we expected to use it in about 200 patients over six months. Instead, we exceeded 1,000 uses in just eight months. The demand was immediate and widespread. Emergency teams used it for post-cardiac arrest patients who weren’t waking up. Neuro ICU teams assessed unresponsive stroke patients. Hospitalists and intensivists quickly adopted the technology. Why? Because it solved a real problem.
Point-of-care EEG changed the equation. In our system, trained bedside providers can apply the device in minutes without waiting for an EEG technologist. The system we use includes an AI algorithm that provides continuous seizure burden monitoring and alerts when seizure activity is detected. It doesn’t replace neurology, but it gives the frontline team immediate clarity and gives neurologists data to review and act on when needed.
Among our first 1,000 patients, about 40% had non-convulsive seizures, and 5% of those patients were in status epilepticus. Just as important, 60% were not seizing. That clarity changed how we treated patients, allowing us to escalate care when necessary and avoid overtreatment when it wasn’t.
In each scenario, point-of-care EEG didn’t just give us data, it gave us confidence. Confidence to treat. Confidence to hold off. Confidence to make the right decision without waiting for consults or transfers.
From Specialty Tool to Scalable Solution
We didn’t introduce point-of-care EEG as a narrow-use tool – we made it a hospital-wide capability. Because recordings are accessible remotely via the cloud, our neurology team supports all cases from a central location. This allows us to scale EEG access without increasing call burden or adding staff. We still rely on full-montage EEG for complex cases, but now we only escalate when specialist input is truly needed.
Point-of-care EEG is now integrated into our stroke and post-cardiac arrest protocols. It’s used across our ICUs, in step-down units, and in the ED. It has expanded access to real-time brain monitoring at our spoke hospitals that lack full EEG coverage.
One case that stands out involved a man in his 60s who arrived at the emergency department with aphasia and right-sided weakness. Although stroke was suspected, both CT and CTA were negative. However, his mental status was markedly depressed, a mismatch that raised suspicion for seizure activity. Point-of-care EEG was quickly initiated at the bedside. Within minutes, it confirmed status epilepticus. Treatment was started without delay, and the patient soon regained consciousness with full resolution of symptoms. The rapid diagnosis helped the patient avoid intubation and led to timely intervention, changing the course of care during triage.
A New Standard for Seizure Assessment
As a stroke neurologist, I’ve always believed that real-time EEG could transform care. Just as no one would tolerate waiting 12 hours for cardiac telemetry in a suspected myocardial infarction, we shouldn’t tolerate EEG delays in patients with unexplained altered mental status.
AI-enabled point-of-care EEG is changing that. It democratizes access by removing barriers including equipment availability, scheduling delays and off-hour staffing, and puts actionable data in the hands of any trained provider. We’ve seen the results: faster action, more targeted care, and a neurology service line that operates more efficiently across the system. This is not just a tool for neurologists. It’s a systemwide capability that helps detect or rule out seizures at the bedside, wherever that bedside may be.
More than 1,000 patient cases have shown us what’s possible. Point-of-care EEG doesn’t just speed up diagnosis – it empowers clinicians, improves systemwide efficiency, and ultimately helps us deliver better outcomes.
Andrei Alexandrov, MD, is chair of the Department of Neurology at the University of Arizona College of Medicine – Phoenix
- Gururangan, K., et al. (2016) Clinical Neurophysiology. 127(10):3335-3340 ↩︎
- Vespa, P., et al. (2020) Crit Care Med. 48(9):1249-1257 ↩︎
- Quigg, M., et al. (2001) J Clin Neurophysiol. 18(2):162-165 ↩︎
- Gavvala, J., et al. (2014) Epilepsia. 55(11):1864-1871 ↩︎
- Bogli, S.Y., et al. (2023) Epilepsia. 64:2409-2420 ↩︎
- Karki, B., et al. (2024) Epilepsia open, 9(1), 325–332 ↩︎
- Laccheo, I., et al. (2015) Neurocrit Care. 22(2):202-211 ↩︎
