Artificial Intelligence–Assisted Risk Stratification for Congenital Syphilis in Antenatal Care

Background: Congenital syphilis remains a preventable cause of perinatal morbidity and mortality, particularly in low- and middle-income countries. Despite universal antenatal screening policies, cases continue to occur due to delayed diagnosis, inadequate treatment, and poor follow-up. Artificial intelligence (AI) offers a novel approach to integrate multiple maternal risk factors for early prediction of adverse outcomes. Objectives: To develop and evaluate an AI-assisted risk stratification model for predicting congenital syphilis in pregnant women diagnosed with syphilis. Methods: A prospective cohort study was conducted among pregnant women with confirmed syphilis attending antenatal clinics of a tertiary care hospital. Clinical, demographic, serological, and treatment-related variables were collected. Machine learning models including logistic regression, random forest, and gradient boosting were developed and compared. Model performance was assessed using area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Results: Out of 280 enrolled participants, 23 neonates (8.2%) were diagnosed with congenital syphilis. The gradient boosting model demonstrated the highest predictive accuracy (AUC 0.88), significantly outperforming conventional risk assessment. High maternal non-treponemal titers, delayed initiation of treatment, inadequate serological response, and lack of partner treatment were the strongest predictors. Conclusion: AI-assisted risk stratification improves prediction of congenital syphilis and may enable targeted antenatal interventions, optimized follow-up, and reduction of preventable adverse neonatal outcomes.