Characterisation of neonatal cardiac dynamics using ordinal partition network

Area of Science:

• Physiology
• Neonatal Medicine
• Complex Systems Analysis

Background:

• Autonomic nervous system (ANS) maturation begins during gestation and continues postnatally, but its development in newborns remains incompletely understood.
• Current clinical assessments of ANS condition in neonates rely on gestational age, Apgar scores, heart rate, and linear heart rate variability methods.
• Nonlinear data analysis approaches have been underexplored for evaluating ANS maturation in newborns.

Purpose of the Study:

• To propose and validate a novel data-driven methodology for classifying autonomic nervous system (ANS) conditions in newborns using nonlinear time series analysis.
• To apply complex network theory to RR interval time series data from premature and full-term newborns.
• To identify complexity quantifiers capable of discriminating between different ANS maturation states in neonates.

Main Methods:

• Development of a data-driven methodology based on nonlinear time series analysis and complex networks.
• Mapping 74 RR interval time series from premature and full-term newborns to ordinal partition networks.
• Calculation of three complexity quantifiers (permutation, conditional, and global node entropies) using forward and reverse network mappings, varying time lags and embedding dimensions.

Main Results:

• Time asymmetry was detected in the RR interval data of both premature and full-term newborn groups.
• The calculated complexity quantifiers demonstrated the ability to differentiate between the analyzed neonatal groups.
• Conditional and global node entropies proved sensitive in detecting subtle differences between neonates, especially with smaller embedding dimensions (m < 7).

Conclusions:

• The proposed complex network-based nonlinear analysis effectively classifies autonomic nervous system (ANS) conditions in newborns.
• Nonlinear techniques, particularly complexity quantifiers derived from ordinal partition networks, offer valuable insights into ANS maturation.
• This study supports the utility of nonlinear methods for analyzing RR interval time series in neonatal research.