Abstract Rationale Respiratory instability in extremely preterm infants is difficult to detect using conventional monitoring, which primarily captures global parameters rather than regional lung function. Electrical impedance tomography (EIT) provides a non-invasive, real-time method to assess regional ventilation, yet standardized frameworks for automated detection of respiratory instability are lacking. Developing an objective analytical approach to identify instability events could enhance early intervention and improve respiratory management in the neonatal intensive care unit (NICU). Methods EIT data were collected from extremely preterm infants (≤32 weeks’ gestational age) in a Level III NICU. A four-stage analytical pipeline was developed: (1) data preprocessing and validation; (2) calculation of tidal volume impedance (TVi) variability using a 30-point rolling coefficient of variation with individualized thresholds (25th percentile = stable, 75th = event) and filtering for sustained periods (≥30 consecutive points); (3) statistical significance testing using Welch’s t-tests, Cohen’s d, and Benjamini-Hochberg FDR correction, applying dual criteria (p 0.05 and d ≥ 0.8 and ≥ 50% change); and (4) event classification integrating nurse-documented clinical events (stimulated vs self-resolved) with quantitative EIT feature extraction within ± 5-minute windows. Results Fourteen infants were analyzed (median 7,111 data points per patient, IQR 6,096-9,642; total 131,037 points). Each infant exhibited a median of 20 stable and 34 event periods. The most consistently significant EIT parameters were coefficient of variation (horizontal) % (86% of patients) and minute ventilation impedance (86%), followed by functional lung size % (79%) and Silent % (71%). Nondependent silent space % was less informative (21%). Among 24 nurse-documented events (10 stimulated, 14 self-resolved), intervention-requiring events showed significantly greater respiratory variability: coefficient of variation (vertical) % (53.0 vs 46.4, p = 0.005, d = 1.49), TVi (101.5 vs 69.0, p = 0.010, d = 1.43), and end-expiratory lung impedance variability SD (61.1 vs 31.1, p = 0.047, d = 1.60). Conclusion A patient-specific TVi variability framework enables objective, automated detection of sustained respiratory instability in extremely preterm infants. This analytical method quantifies physiologic changes that differentiate intervention-requiring from self-resolved events, offering a foundation for personalized respiratory monitoring and future decision-support tools in the NICU. This abstract is funded by: None
Leibel et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: