Skip to main content
SMFM 2025 Pregnancy Meeting Main banner
Final Program


Program Book
Icon Legend
Presentation Icons
Additional registration fee required
Additional registration fee required
CME Credit Offered
CME Credit Offered
Abstract Award
Abstract Award
Poster Icons
Additional registration fee required
Additional registration fee required
Faculty have requested this content not be shared on social media
Faculty have requested this content not be shared on social media
CME Credit Offered
CME Credit Offered
Abstract Award
Abstract Award
Back

Poster Session 1

(108) Leveraging Machine Learning Models to Identify Predictors of RSV Immunization During the 2023-24 Season

Thursday, January 30, 2025
10:30 AM – 12:30 PM  
Objective: In 2023, the FDA approved maternal RSVpreF vaccines and infant nirsevimab immunization for prevention of infant respiratory syncytial virus (RSV) disease. This study aims to use unsupervised machine learning to identify key predictors of RSV immunization.

Study Design: Maternal-infant dyad demographic and clinical data was obtained from the electronic medical record. RSV immunization status was abstracted from a state-wide database. The outcome variables were RSV immunization status (maternal, infant, dual, or unvaccinated). Predictors were maternal age, gestational age at birth, delivery month, obstetric provider group, payor category, parity, self-reported race, language, perinatal demise, infant number, and sex. Eleven machine learning classification models (logistic regression, LASSO, RIDGE, Random Forest, SVM Radial and Linear, XGBoost, BNB, MLP, KNN, Catboost) were compared using the AUC-ROC One vs Rest (OvR) score. SHAP analysis was employed to identify the most important features.

Results: 1899 births (1940 live born infants) from September 2023 to January 2024 were included. The Catboost model demonstrated the best performance (AUC-ROC OvR score: 0.75). Maternal age, gestational age, provider group, delivery month, and language ranked highest in SHAP summary plots (Figure 1). Higher maternal age and gestational age were associated with maternal vaccination; lower maternal and gestational age were associated with infant immunization and unvaccinated status. Maternal vaccination was more likely later in the season. Private practice was associated with maternal/dual vaccination, while community or academic practices were associated with infant immunization. English speakers were more likely to receive maternal vaccination, whereas non-English speakers were more likely to receive newborn or no immunization.

Conclusion: Results from this model suggest that increasing vaccine availability in all practice settings throughout the season and providing language-concordant education could improve equitable implementation of RSV immunization.