Assessment of heart rate variability by short-time Fourier transform and data analysis

Heart rate variability (HRV) analysis of 24 healthy newborn sleeping babies was performed by a short-time Fourier transform in three frequency bands, reflecting the activity of both branches of the autonomic nervous system (ANS), vagal and sympathetic. The means of the three extracted time signals, computed over records of 512 heartbeats, were used as a material for principal component analysis, and for discriminant factor analyses, to separate sleep states and conceptional age (CA) groups. This study suggests that sleep state discrimination, on the basis of an opposition between high (purely vagal in its origin) and low (vagal and sympathetic) frequency HRV, is regularly improved from 31 to 41 weeks CA; and a strong increase in ANS activity, mainly vagal, as reflected by high-frequency HRV, occurs precociously, not later than 38 weeks CA. >