Point Process Analysis of Two Alternating Physiological Events: Application to Movement Bouts of Preterm Infants

A temporal point process is a statistical framework that is used to model a continuous process with discrete measurable events. Here we apply point process theory to the special case of two alternating events. Time series with alternating events are commonly measured in physiological systems, for example, alternation between inspiration and expiration, systole and diastole, wakefulness and sleep. As our illustrative example, we develop and apply point process theory to characterize the alternation between the onset and offset of each movement bout exhibited by preterm infants in the neonatal intensive care unit (NICU). Our point process model captures the probabilistic dynamics of movement bouts with favorable goodness-of-fit and reveals that previous onset events have an excitatory effect on the next onset, and previous offset events have an inhibitory effect on the next offset. The offset events are modulated by recent onset events. These findings provide a quantitative framework for understanding the dynamics of preterm infant movements as a function of prior movements. Future work could extend this model to investigate how these movement characteristics evolve with healthy maturation and disease states associated with adverse neurological outcomes. We suggest that a multivariate point process framework can provide insights into physiological processes with two or more alternating measured events.