Related Experiment Video
Updated: Apr 17, 2026
11:26
Real-time X-ray Imaging of Lung Fluid Volumes in Neonatal Mouse Lung
Published on: July 18, 2016
9.3K
An algorithm to identify less invasive surfactant administration using a real-world database of preterm infants
Xuezheng Sun1, Annie N Simpson2, Aditi Lahiri3
1Chiesi USA Inc, Cary, North Carolina, United States of America.
Plos One
|April 15, 2026
Summary
A new algorithm accurately identifies less invasive surfactant administration (LISA) in preterm infants using administrative data. This enables large-scale research into LISA
Area of Science:
- Neonatal Medicine
- Health Informatics
- Data Science
Background:
- Respiratory distress syndrome (RDS) management in preterm infants relies on surfactant replacement therapy.
- Less invasive surfactant administration (LISA) improves neonatal outcomes but lacks administrative procedure codes.
- This limits large-scale evaluation of LISA in real-world data (RWD).
Purpose of the Study:
- To develop and validate an algorithm for identifying LISA procedures using administrative data.
- To facilitate large-scale analysis of LISA utilization and effectiveness.
Main Methods:
- Retrospective study using chart reviews as the gold standard.
- Developed a LASSO regression model with 21 variables from administrative data (2019-2023).
- Validated the algorithm using a combined testing set and 2024 birth cohort, assessing sensitivity, specificity, PPV, and NPV across gestational age subgroups.
Main Results:
- The algorithm achieved an AUROC of 0.87, indicating strong discrimination.
- At a specificity cut-point (≥0.79), the model yielded sensitivity 43.9%, specificity 96.8%, PPV 90.0%, and NPV 72.5%.
- Overall agreement was 75.9%, with consistent performance across gestational age subgroups.
Conclusions:
- A machine-learning algorithm effectively identifies LISA in preterm infants using administrative data.
- The validated algorithm supports future research on LISA utilization and outcomes in RWD.
