Developing a model to predict neonatal respiratory distress syndrome and affecting factors using data mining: A...

Area of Science:

• Medical Informatics
• Neonatal Medicine
• Machine Learning

Background:

• Early identification of neonatal adverse events, including respiratory distress syndrome (RDS), is a significant clinical challenge.
• RDS is a prevalent respiratory disorder in premature infants, contributing to mortality.
• Machine learning offers valuable tools for analyzing medical data and detecting RDS early.

Purpose of the Study:

• To develop a predictive model for neonatal RDS using data mining techniques.
• To identify factors influencing the occurrence of neonatal RDS.
• To evaluate the performance of various machine learning algorithms for RDS prediction.

Main Methods:

• A cross-sectional study utilizing a dataset of 1469 neonates and their mothers from Alzahra hospital, Iran (July 2017-July 2018).
• Data preprocessing followed by the application of machine learning algorithms: Support Vector Machine, Naïve Bayes, Classification Tree, Random Forest, CN2 Rule Induction, and Neural Network.
• Model performance was compared based on accuracy, sensitivity, specificity, and Area Under the Curve (AUC).

Main Results:

• The Random Forest model demonstrated the best performance with an accuracy of 0.815, sensitivity of 0.802, specificity of 0.812, and AUC of 0.843.
• Comparison of various machine learning techniques was conducted to identify the most effective for RDS prediction.

Conclusions:

• Data mining approaches, particularly Random Forest, show promise in supporting clinical decisions for improved neonatal RDS diagnosis.
• The feasibility of using Random Forest for neonatal RDS prediction can potentially decrease postpartum complications in neonatal care.