Preterm infants in this group will receive routine breast milk feeding according to standard NICU practice. Expressed breast milk will be prepared and administered without matching the time of milk expression to the time of feeding.

Preterm infants in this group will receive circadian-synchronized breast milk feeding in the NICU. Expressed breast milk will be classified and labeled according to the time of milk expression into 4 time periods: morning (06:00-12:00), afternoon (12:00-18:00), evening (18:00-24:00), and late night/early morning (00:00-06:00). During the study, breast milk collected within each predefined time period will be used preferentially for feedings scheduled within the corresponding time period. This feeding approach is intended to keep the timing information in breast milk as consistent as possible with the infant's feeding time and will continue for at least 2 weeks during the study.

In this intervention, expressed breast milk will be classified and labeled according to the time of milk expression into 4 predefined periods: morning (06:00-12:00), afternoon (12:00-18:00), evening (18:00-24:00), and late night/early morning (00:00-06:00). During the study, breast milk collected within each predefined time period will be used preferentially for feedings scheduled within the corresponding time period. This intervention is designed to keep the timing information in breast milk as consistent as possible with the infant's feeding time and will be implemented for at least 2 weeks.

The primary outcome is the ratio of active sleep (AS) to quiet sleep (QS) across the entire monitoring period at 37 weeks of corrected gestational age. Active sleep is one of the 2 main neonatal sleep states and is considered the precursor of rapid eye movement (REM) sleep. Sleep data will be collected during a 12-hour monitoring period using a wearable EEG patch combined with a sleep mat. The EEG device uses a single-channel, three-electrode configuration and supports synchronized acquisition of EEG and pulse wave signals. All monitoring data will be preprocessed, scored, and exported using a standardized data processing workflow.

The proportion of time spent in each sleep state during the entire monitoring period, including active sleep (AS), quiet sleep (QS), intermediate sleep (IS), arousal, and wake. Each sleep state will be expressed as a percentage of the total monitoring time.

Stool samples will be collected at enrollment and on the same day as sleep monitoring at 37 weeks of corrected gestational age to explore gut microbiota composition, diversity, and relative abundance of major bacterial taxa. Analyses will be exploratory.