麻豆果冻传媒影音

DENVER -- A high seizure burden derived by an artificial intelligence (AI) algorithm was associated with worse outcomes, a retrospective analysis showed.

Hospitalized patients who had a high seizure burden based on a point-of-care electroencephalography (EEG) algorithm had four times higher odds of an unfavorable functional outcome compared with those who had no seizure burden, reported Masoom Desai, MD, of the University of New Mexico in Albuquerque.

The algorithm's assessment of seizure burden also predicted hospital discharge to long-term care facilities, she said at the American Academy of Neurology (AAN) annual meeting.

"Several decades of research highlighted the significant correlation between seizure burden and unfavorable outcomes in patients with various neurological conditions, both in adult and pediatric populations," Desai said during an AAN press conference.

"The traditional method of manually interpreting EEGs to identify seizures and their associated burden is a complex and time-consuming process that can be subject to human error and variability," she continued. "Our study addresses the critical need for automation in monitoring epileptiform activity and seizure burden."

The study is the first to use an automated algorithm to correlate sustained point-of-care seizure activity and functional outcomes, she noted.

Continuous EEG recordings of critically ill patients commonly show epileptiform activity associated with worse patient outcomes, including disability and death. A recent retrospective study using continuous EEG demonstrated that was correlated with neurologic outcome even after adjusting for confounding variables like antiseizure medications.

However, continuous EEG is not widely available, Desai pointed out. Clarity is a machine-learning algorithm cleared by the FDA that about suspected status epilepticus using point-of-care EEG recordings, she said.

The study was a sub-analysis of data from the retrospective SAFER-EEG trial. The Clarity algorithm was run post-hoc in 344 cases with point-of-care EEGs at four academic centers.

Desai and colleagues analyzed outcomes in patients with a maximum seizure burden of 50% or more near the end of the recording, including those with a suspected status epilepticus alarm (a seizure burden of 90% or more). They defined an unfavorable functional outcome as a modified Rankin scale (mRS) score of 4 or higher. Scores on the mRS range from 0 to 6; a score of 4 indicates the patient is unable to walk or attend to bodily functions without assistance.

Of the 344 cases, 178 people (52%) had a seizure burden of 0 throughout the recording, and 41 (12%) had suspected status epilepticus.

A total of 31 patients had a seizure burden of 50% or more in the last quarter of the recording. These patients were more likely to have unfavorable mRS outcomes compared with patients who did not have any seizure activity (76% vs 57%, P=0.035). "A similar proportion was discharged to long-term care facilities," Desai noted.

After adjusting for clinical covariates, the researchers found that high seizure burden raised the odds of unfavorable mRS outcomes at hospital discharge (OR 4.1, 95% CI 1.8-9.6, P≤0.01).

The seizure burden in the last quarter of the recording was particularly indicative of outcomes, "suggesting the critical timing of seizures and its impact on patient prognosis," Desai observed.

An AI algorithm may be able to help guide more effective treatment strategies and possibly mitigate some adverse effects of prolonged seizure activity on patient outcomes, Desai suggested. More research is needed, and the next step is to prospectively assess the potential of EEG machine-learning algorithms, she added.

"As we move forward, our research will concentrate on applying this advanced tool in clinical decision making and clinical practice, examining how it can steer treatment decisions for patients," she said.

Comment