Abstract
Computational analysis of infant movement has significant potential to reveal markers of developmental health. We report two studies employing dynamic analyses of motor kinematics and motor behaviours, which characterise movement at two levels, in 9-month-old infants. We investigate the effect of preterm birth (< 33 weeks of gestation) and the effect of changing emotional and social-interactive contexts in the still-face paradigm. First, multiscale permutation entropy was employed to analyse acceleration kinematic timeseries data collected from Inertial Measurement Unit (IMU) sensors on infants' torso, wrists, and ankles (N = 32: 10 term; 22 preterm). Second, Recurrence Quantification Analysis was used to characterise patterns of second-to-second behavioural changes, from observationally coded behavioural timeseries on infants' emotional self-regulation (N = 111: 61 term; 50 preterm). We found frequency-specific effects of context on permutation entropy. Relative to infants born at term (> 37 weeks of gestation), infants born preterm showed greater permutation entropy in their left ankle and torso movements, but not in right ankle or wrist movements. We did not find effects of preterm birth or emotional context on micro-level behavioural dynamics. Our methodology and findings inform future work using multiscale entropy to study infant development. Dynamic analysis of behaviour is a relatively young field, and applications to emotional self-regulation requires further methodological development.