Using of local information for diagnosing intranatal adverse conditions by pulse oximetry

The study aims to use the predictive value of oxygen saturation for the determination of hypoxemia and acidosis by local information during intranatal fetal monitoring. For this purpose, we employ the measurement of umbilical artery (UA) and umbilical vein (UV) oxygen saturation and pH in an unselected population, and calculate preductal oxygen saturation at birth. The study analyzes umbilical cord blood samples of 1537 live-born singleton neonates. Oxygen saturation, pH and base excess were measured. Preductal oxygen saturation was computed with an empirical equation. Acidosis was defined as below the value of 7.09 for UA pH or -10.50 mmol/L for base excess. The normality condition of the data is searched by the Kolmogorov-Smirnov test. Then, the importance of individual indices is determined by using principal component analysis (PCA) method. A local decision technique, k-nearest neighbor (k-NN) method is employed for the indication of adverse conditions. Finally, RBF is used to combine the local information regions. It is verified that the base excess in UA is the most informative component and the local oxygen saturation information is valuable for the indication of adverse conditions.