/Cesareandelivery-prediction

This study aimed to assess whether machine learning algorithms would outperform traditional modeling in developing a cesarean delivery prediction model among gravidas with morbid obesity (body mass index of ≥40 kg/m2) to determine whether a primary cesarean delivery may be beneficial.

Primary LanguageJupyter Notebook

Rajasri

DSA Final Semester

Cesarean delivery prediction

A retrospective cohort of 1287 women delivering at one institution from 2011-2016 with 42 available demographic and clinical variables were examined using random forest recursive feature elimination (RF-RFE) and support vector machine recursive feature elimination (SVM-RFE) to select variables for analysis.

Subsequently, logistic regression (LR) was compared to machine learning algorithms including decision tree (DT), random forest (RF), support vector machine with the linear kernel (SVM-Lin), and support vector machine with the radial basis function kernel (SVM-RBF) in their ability to predict cesarean as a binary outcome.

70% of the cohort was used to train the algorithm and 30% was used for validation, for a total of 50 iterations. The area under the receiver operator curve (AUC) was used to determine the ability of an algorithm to correctly classify the outcome and p< 0.05 was considered significant.

SVM-RFE was more inclusive compared to RF-RFE with regards to feature selection and determined 22 significant variables compared to 10 (p < 0.001).