Related Experiment Video
Updated: Jun 12, 2025
Early Pathological and Magnetic Resonance Detection of Cerebral Injury Using a Rat Model of Neonatal Hypoxic Ischemic Encephalopathy
Published on: October 28, 2022
Development and Validation of a Neonatal Hypothermia Prediction Model for In-Hospital Transport Using Machine
Wenyan Zhang1, Xiaoying Gu1, Chunjie Gu1
1Department of Neonatology, Shanghai Key Laboratory of Maternal Fetal Medicine, Shanghai Institute of Maternal-Fetal Medicine and Gynecologic Oncology, Shanghai First Maternity and Infant Hospital, School of Medicine, Tongji University, Shanghai, 200092, People's Republic of China.
Neonatal hypothermia during hospital transport is common. A Random Forest machine learning model effectively predicts hypothermia risk, identifying key factors like gestational age and weight for early intervention.
Area of Science:
- Neonatal Medicine
- Machine Learning in Healthcare
- Clinical Informatics
Background:
- Neonatal hypothermia during in-hospital transport presents a significant clinical challenge.
- Early identification and intervention are crucial for improving neonatal outcomes.
Purpose of the Study:
- To develop and validate a machine learning model for predicting hypothermia in neonates during in-hospital transport.
- To identify and rank key risk factors associated with neonatal hypothermia during transport.
Main Methods:
- Utilized clinical data from 9,060 neonates, employing LASSO regression for variable selection.
- Trained and evaluated six machine learning algorithms (DT, RF, XGBoost, SVM, ANN, NB) on transport temperature data.
- Assessed model performance using AUC, F1 score, accuracy, sensitivity, specificity, and calibration tests.
Main Results:
- Over 55% of neonates experienced hypothermia during transport.
- The Random Forest (RF) model achieved superior performance with a test set AUC of 0.962 and accuracy of 0.889.
- Identified ten significant risk factors, including gestational age, weight, and immediate postnatal contact.
Conclusions:
- High incidence of hypothermia underscores the need for predictive tools.
- The RF-based model offers robust predictive and generalization capabilities for identifying at-risk neonates.
- Provides actionable insights for healthcare providers to mitigate hypothermia risk during neonatal transport.
