Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
June 13, 2026ChildrenOpen Access

Developing and Validating a Machine Learning Model to Predict Brain Injury in Preterm Infants Using Multisource Data from the Early Postnatal Period

View Full Paper
Ask AI
Bookmark
Share

Authors

PXPu XuYLYing LiYCYing Chen

Discussion

Loading...

Member takes

Overview

Randomized trial develops a machine learning model to predict brain injury in preterm infants, indicating potential clinical applications.

Key Points

  • The aim is to develop and validate a machine learning model to predict brain injury in preterm infants using early postnatal data.
  • Retrospective study of 318 preterm infants admitted from 2015 to 2024.
  • Evaluation of 33 candidate predictors from clinical and laboratory data, implemented through machine learning.
  • Model selection and calibration using LightGBM with cross-validation and performance metrics such as AUROC.
  • PBI occurred in 19.5% of the development cohort (62/318) and 17.1% of the external cohort (6/35).
  • The finalized LightGBM model achieved an AUROC of 0.747 (95% CI 0.679–0.811) after Platt calibration.
  • In the temporal external cohort, the AUROC was 0.897 (95% CI 0.672–1.000), but results are considered preliminary due to the small sample size.

Cite This Study

Xu et al. (2026) studied this question.

synapsesocial.com/papers/6a2cf488faef96ed7f056bb3https://doi.org/10.3390/children13060796
View Full Paper
Ask AI
Bookmark
Share