Introduction

Continuous cardiotocography (CTG) is the most widely used method of intrapartum fetal surveillance, but its clinical performance is constrained by substantial interobserver disagreement, even under standardized three-tier nomenclature, and by low specificity for clinically important acidemia [1-3]. These limitations have direct clinical consequences: poor specificity may contribute to unnecessary cesarean and operative vaginal deliveries and their attendant maternal morbidity, while limited sensitivity may allow some compromised fetuses to escape timely intervention. Computerized interpretation has not consistently improved neonatal outcomes, and artificial-intelligence approaches remain heterogeneous in design, transparency, and validation [4,5]. There is therefore a need to investigate interpretable and reproducible features that might eventually complement conventional visual CTG assessment. Human studies suggest that fetal acidemia may be accompanied by redistribution of heart-rate variability across frequencies rather than by a simple change in overall amplitude [6-8]. The Fourier Evaluation of Tracings and Acidosis in Labor (FETAL) framework asks whether the normalized Fourier power spectrum of a test tracing differs from a reference spectrum and returns a transparent spectral-divergence feature intended for later evaluation within an intrapartum clinical prediction model. The present study does not determine whether this feature improves prediction beyond established CTG variables, computerized CTG features, or clinician interpretation. No conclusions regarding incremental predictive value, clinical diagnostic accuracy, or clinical benefit can be drawn from an entirely synthetic experiment. Before those questions can be evaluated in patient-level data, however, the computational construct must first be shown to behave as intended and, critically, to remain uninformative when no detectable spectral difference is present. In this report, we sought to accomplish two main objectives. The primary objective was to test three prespecified properties of the spectral-divergence score: chance discrimination when no class difference exists, increasing discrimination as the imposed within-band spectral shift strengthens, and reduced discrimination under artifact stress. The secondary objective was to compare the full-spectrum score with overall standard deviation and the low-/high-frequency power ratio. These objectives concern computational identifiability and falsifiability, not clinical validity, incremental predictive performance, or a causal association between spectral divergence and fetal acidemia.