Machine Learning Algorithms as a Tool to Study Hospitalization for Cesarean Section: A Multicenter Study

Cesarean sections (CS) are a surgical procedure of strong interest in health research. While strategies are being implemented to reduce the use of this procedure especially in primiparas without clinical indication, efforts are also being made to make the process increasingly optimized and standardized. This study conducted at the Evangelical Hospital "Betania" in Naples, Italy, aims to use simple Machine Learning algorithms to predict and evaluate hospital stay for CS. The analysis traces what has already been published for 3 other hospitals in southern Italy that together constitute a significant share of the birth points in the Campania region. An interesting fact is that for all the hospitals studied Random Forest is the best classifier, stopping in this study at 81% slightly worse than the value obtained with data from AOU Ruggi and AORN Cardarelli. However, the result remains optimal such that it validates the use of this algorithm for the study of hospitalization.