An early prediction model to identify neurological complications of
childhood influenza: a random forest model
Abstract
Background: NeurologicalEarly prediction model for seizures in influenza
complications of influenza are associated with high morbidity and
mortality in children. The prognosis could be improved if early
treatments are undertaken. Objective: To establish and validate an early
prediction model to discriminate among neurological complications such
as seizures, acute influenza virus-associated encephalitis (IAE), and
acute necrotizing encephalopathy (ANE) in children with influenza.
Methods: This was a retrospective single-center case-control study
conducted at the Guangzhou Women and Children’s Medical Center in
Guangzhou (GWCMC), China, from November 2012 to January 2020. The random
forest model was used to screen the characteristics and construct an
early prediction model for convulsions, IAE, and ANE. Results: Of the
433 patients (294 male, 139 female; median age 2.8 (1.7,4.8) years), 278
(64.2%) had seizures, 106 (24.5%) had IAE, and 49 (11.3%) had ANE;
348 patients were in the training set and 85 in the validation set. When
10 variables were included, the cross-validation error was minimal;
convulsions, procalcitonin, urea, γ-glutamyltransferase, aspartate
aminotransferase, albumin/globulin ratio, α-hydroxybutyric
dehydrogenase, alanine aminotransferase, alkaline phosphatase, and
C-reactive protein were included. The likelihood of having only seizures
decreased with increasing procalcitonin, urea, γ-glutamyltransferase,
α-hydroxybutyric dehydrogenase, alanine aminotransferase, and aspartate
aminotransferase, and with decreasing albumin/globulin ratio and
alkaline phosphatase. The prediction model gave a prediction accuracy of
84.2%. Conclusion: This model can distinguish the seizures from IAE and
from ANE. This could allow for the early management of children with
influenza in order to prevent morbidity and mortality. The
biochemical/hematologic markers lacked specificity.