Caesarean Section Prediction Using Machine Learning with DALEX Explainability

Le résumé fourni par la source

Caesarean section (C-section) is an crucial surgical procedure that, when clinically warranted, can save maternal and neonatal lives.But its increasing worldwide prevalence underlines the urgent need for reliable and interpretable decision-support system. In this study, we have created an AI driven Clinical Decision Support System (CDSS) which utilizes Machine Learning (ML) and Explainable Artificial Intelligence (XAI) to increase the transparency and accuracy of C-section prediction. We have applied four widely used ML classifiers, Multi-Layer Perceptron (MLP), Decision Tree (DT), Random Forest (RF) and Extreme Gradient Boosting (XGBoost) inorder to determine whether a C-section delivery is needed based on patient attributes. Standard classification metrics are used to evaluate the models, such as accuracy, precision, recall (sensitivity), specificity, F1-score, and Geometric Mean (GM) under a 10 fold cross validation scheme. Among the evaluated models ensemble based approaches like XGBoost and RF demonstrate superior predictive performance. Descriptive ML EXplanations (DALEX) framework is employed to interpret global and local feature contributions in order to ensure model transparency and promote clinical adoption. DALEX identifies Delivery Time, Age, Heart Problem, and Blood Pressure as key predictors influencing C-section decisions. The proposed interpretable ML framework offers enhanced performance while maintaining clinical interpretability, making it a promising tool for practical obstetric decision-making.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.