Predicting spontaneous preterm birth using machine learning algorithms.

Area of Science:

• Obstetrics and Gynecology
• Medical Imaging
• Machine Learning in Healthcare

Background:

• Spontaneous preterm birth (sPTB) is a leading cause of neonatal morbidity and mortality.
• Early identification of at-risk pregnancies is crucial for timely intervention.
• Cervical elastography offers a non-invasive method to assess cervical biomechanics.

Purpose of the Study:

• To evaluate the predictive value of cervical elastography parameters for sPTB using machine learning.
• To identify key predictors of sPTB in a cohort of pregnant women.
• To develop accurate ML-based models for sPTB risk stratification.

Main Methods:

• Prospective cohort study of pregnant women (2022-2023).
• Analysis of demographic data, clinical history, and cervical elastography parameters.
• Application of four ML algorithms (Boruta, Lasso, SWSFS, XGBoost) for predictor screening and model construction.

Main Results:

• The incidence of sPTB was 9.06%.
• Cervical length (AUC=0.784), anterior cervical angle (AUC=0.731), posterior cervical angle (AUC=0.623), and combined angles (AUC=0.674) were significantly associated with sPTB.
• The Boruta-based logistic regression model achieved the highest predictive performance (AUC=0.953).

Conclusions:

• Cervical length ≤3.015 cm, combined cervical angles >204.49°, prior sPTB, and abnormal strain rate are critical predictors of sPTB.
• ML models demonstrate high accuracy for predicting sPTB.
• These models show promise for early clinical risk stratification, particularly in ethnic minority populations.