Neonatal blood glucose monitoring using glucose estimation by phase delay between oxy- and deoxyhemoglobin

Reliable, painless glucose surveillance is critical during the early neonatal fall in plasma glucose levels. However, existing continuous glucose monitors are invasive and expensive. We evaluated a non-invasive method for estimating blood glucose levels from the phase delay (Δθ) between oxy- and deoxyhemoglobin waveforms, expressed as a metabolic index (MI).

Thirty-eight term neonates were enrolled on postnatal day 2, and 30 recordings satisfied the predefined signal quality criteria. Neonatal pulse oximetry probes captured 5-min photoplethysmography traces, from which stable 15-60-s epochs were extracted. The α-corrected MI for each infant was correlated with plasma glucose measured on an ABL90 FLEX analyzer. Glycemic slopes were compared with published adult data using ANCOVA.

The α-corrected MI showed a significant positive correlation with plasma glucose (R² = 0.51, r = 0.71; p < 0.01). Linear regression for neonates was MI = 0.17 × glucose - 6.78. Both slope and intercept differed from adult values (p < 0.05), indicating age-dependent modulation of the Δθ-glucose relationship.

Phase-delay analysis using standard pulse oximetry provides a promising approach for continuous glucose estimation in term neonates. Age-specific calibration is required before clinical deployment; however, the technique's low cost and noninvasiveness potentiate universal bedside hypoglycemia surveillance.

Key message: Non-invasive glucose monitoring using the phase delay between oxy- and deoxyhemoglobin waveforms obtained from a pulse oximetry probe shows a significant correlation with blood glucose levels in term neonates. Literature contribution: This study demonstrates that the metabolic index derived from hemoglobin phase delays differs between neonates and adults, suggesting the need for age-specific calibration.

This low-cost technique may enable continuous bedside glucose surveillance in neonates without repeated blood sampling.