Prediction of the mode of delivery using artificial intelligence algorithms

Area of Science:

• Obstetrics and Gynecology
• Artificial Intelligence in Medicine
• Clinical Decision Support Systems

Background:

• Rising caesarean section rates exceed recommendations.
• Obstetricians need better tools for delivery mode decisions.
• Antepartum and intrapartum conditions influence delivery outcomes.

Purpose of the Study:

• To evaluate artificial intelligence (AI) algorithms for predicting delivery mode.
• To develop a clinical decision support system for obstetricians.
• To classify deliveries into caesarean, euthocic vaginal, and instrumental vaginal categories.

Main Methods:

• Utilized Support Vector Machines, Multilayer Perceptron, and Random Forest AI algorithms.
• Developed a clinical decision support system using a database of 25,038 patient records.
• Analyzed data from singleton pregnancies from January 2016 to January 2019.

Main Results:

• All three AI algorithms demonstrated high performance.
• Achieved accuracy above 90% for caesarean vs. vaginal delivery classification.
• Attained approximately 87% accuracy for instrumental vs. euthocic vaginal delivery classification.

Conclusions:

• Validated the use of AI algorithms for predicting delivery mode.
• Results support the development of AI-driven clinical decision systems.
• Aids gynecologists in making informed decisions regarding delivery methods.