Pattern discrimination software for neonatal cardiovascular risk estimation

Abstract

Parametric characteristics obtained by least-squares fitting and descriptive statistics of within-a-day variability in blood pressure and heart rate for 150 neonates are here used for classification according to a cardiovascular risk score by the use of an all-subsets variable selection technique for discriminant analysis, specifically designed for biomedical applications. Results provide a set of classification criteria for further use in neonatal risk assessment.1