Lempel-Ziv complexity of the pNNx statistics – an application to neonatal stress

Abstract Among the existing measures of heart rate variability (HRV), the pNN50 statistics is one of the most commonly reported. However, it is only a single member of a much larger family of HRV measures – the pNN x statistics. In this research pNN x was further extended, combining it with the Lempel-Ziv complexity (LZ76) in a controlled neonatal stress framework. Two different types of stress stimuli on forty healthy newborns – a routine heel stick blood sampling, and a dull heel pressure stimulation – were considered by recording time intervals between heartbeats. Instead of relying on a single value, the entire spectrum from pNN1 to pNN100 was calculated, along with LZ76 derived from binarized sequences for each NN. The results of this study show a downward shift of the pNN x curves when newborns are stressed, with reduced LZ76 complexity when stressed. When ROC curves were utilized for the pNN x statistics and LZ76, however, the highest AUC values were observed when both measures were combined, with the highest AUC values of 0.88 (0.80-0.94) and 0.85 (0.74-0.91) for discriminating resting states from stress phases. Combining the widely used pNN x statistics with LZ76 extends the existing HRV toolbox, and shows a promising application in recognizing acute neonatal stress.