Early Maturation of Heart Rate Variability in Very Preterm Infants Depends on Neonatal Factors and Is Associated With Neurodevelopmental Risk

FIGURE 1. Overview of the study. (a) The whole process for signal processing, HRV feature extraction and selection, and Ensemble Machine Learning used has already been described in detail [16]. (b) ΔHRV was defined as the difference between post‐menstrual age (PMA) and functional autonomic maturation age estimated by machine learning (FMA). The dashed line corresponds to the expected ideal central autonomic network maturation built in the training phase of the study. The observed individual FMA trajectories may have been influenced by prematurity and perinatal events. (c) Using the parental questionnaire Ages and Stages Questionnaire, third edition, we assessed if ΔHRV was related to perinatal events and to neurodevelopmental outcome at the corrected age of 2 years.

FIGURE 2. Points represent individual functional maturation age (FMA) estimated from birth to hospital discharge. Lines represent the median FMA. The results are presented according to the presence (blue) or absence (red) of complications: (a) bronchopulmonary dysplasia; (b) patent ductus arteriosus; c. neurological lesions.

FIGURE 3. Points represent individual functional maturation age (FMA) estimated from birth to hospital discharge. Lines represent the median FMA. The results are presented based on the results of personal‐social skill assessments (Ages and Stages Questionnaire) at 2 years: In blue for preterm infants with impaired personal‐social skills (score more than two standard deviations below the expected for age) in red for the other preterm infants.