Pointwise Hypothesis Testing of Biomedical Near-Infrared Spectroscopy Signals

This study uses a pointwise statistical approach to analyze Near-Infrared Spectroscopy (NIRS) signals in preterm infants with and without Patent Ductus Arteriosus (PDA). The analysis focuses on three signals: blood oxygenation (SpO2), cerebral oxygenation (rSO2-1), and renal oxygenation (rSO2-2), across three newborn groups: without PDA (no-PDA), with hemodynamically insignificant PDA (PDA), and with hemodynamically significant PDA (hsPDA). While NIRS is widely used in medicine, its research, featuring statistical analysis, has been limited. Smoothed signals were tested using pointwise ANOVA and Tukey HSD to detect significant group differences. Results showed distinct patterns in rSO2-1 and rSO2-2, with the hsPDA group standing out in rSO2-1 and the no-PDA group in rSO2-2, demonstrating the value of this method in biomedical signal analysis. Pointwise ANOVA shows more time periods with significant differences compared to the SpO2 signal. The time period with the most significant differences is between 2 and 6 h, with additional peaks of p-values below 0.05 occurring before 2 h. These findings demonstrate the value of FDA in improving statistical analysis of biomedical NIRS signals and support its use in future research.