Diagnostic methodology for the neonatal cardiac system based on dynamic systems

Abstract

Fractal geometry and nonlinear systems have shown to be useful to diagnose and predict the behavior of cardiac dynamics in adults with promising applications, but studies in neonates are lacking. The objective of this study is to apply a chaotic methodology to diagnose heart dynamics of neonates. 70 electrocardiographic records were taken from newborn patients, 10 normal and 60 pathological. Then, the minimum and maximum heart rates as well as the number of heartbeats/hour were taken for at least 21 hours; a sequence of heart rates was generated and, based on that sequence, chaotic attractors were constructed. Their fractal dimension was calculated as well as their occupation in the Box-Counting space and the mathematical diagnosis was determined based on the limits established in the induction. A blind study was developed for statistical validation against the Gold Standard diagnosis. It was found that the spatial occupation of neonatal cardiac chaotic attractors in the Box-Counting fractal space allowed the differentiation between normality and disease, reaching the highest values of sensitivity, specificity and diagnostic concordance against Gold Standard. Hence, it is concluded that fractal geometry allows to diagnose and differentiate normal from abnormal heart dynamics of neonates independent of the clinical scenarios, corroborating its utility at clinical level.