Data Extracted From Continuous EEG Can Predict Neonatal Seizures

Data extracted on first day of continuous EEG can speculate presence of neonatal seizures on following days with high accuracy.

HealthDay News Data extracted from standardized electroencephalogram (EEG) on the first day of continuous EEG (CEEG) monitoring can help predict neonatal seizures on subsequent days, according to a study published in the April issue of The Lancet Digital Health.

Jillian L. McKee, M.D., from the Children’s Hospital of Philadelphia, and colleagues conducted a retrospective cohort study of neonates who underwent CEEG during the first 30 days of life to generate seizure prediction models. Logistic regression, decision tree, and random forest models of neonatal seizure prediction using EEG features on day 1 were developed to predict seizures on future days. A total of 1,117 neonates were evaluated, including 150 with hypoxic ischemic encephalopathy.

Researchers found that several EEG features were highly correlated, and on the basis of specific features, patients could be clustered. However, seizure risk was not adequately predicted by any simple combination of features. Computational models were applied to complement clinical identification of neonates at high risk for seizures. Classification accuracies were up to 90% for all neonates and 97% for neonates with hypoxic ischemic encephalopathy for random forest models incorporating background features; recall (sensitivity) was up to 97 and 100% for all neonates and for neonates with hypoxic ischemic encephalopathy; and precision (positive predictive value) was up to 92 and 97%, respectively.

“If validated, this model could enable more targeted use of limited CEEG resources by reducing CEEG duration among patients at low risk of seizures after the initial day of CEEGs,” the authors write.

Abstract/Full Text