AIDA (Artificial Intelligence Dystocia Algorithm) in Prolonged Dystocic Labor: Focus on Asynclitism Degree

Area of Science:

• Obstetrics and Gynecology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Asynclitism, a fetal head misalignment, is a significant obstetric challenge associated with difficult labor and increased cesarean delivery rates.
• Accurate diagnosis and understanding of asynclitism's impact on labor outcomes remain debated.
• Intrapartum ultrasound (IU) offers objective measurements for assessing fetal head positioning.

Purpose of the Study:

• To analyze the role of the degree of asynclitism (AD) in predicting labor progress and intrapartum cesarean delivery (ICD) versus non-cesarean delivery.
• To evaluate the performance of the Artificial Intelligence Dystocia Algorithm (AIDA) in integrating AD with other ultrasound parameters for labor outcome prediction.
• To assess the predictive accuracy of the AIDA algorithm using various machine learning models.

Main Methods:

• Retrospective study of 135 nulliparous patients with singleton cephalic presentations undergoing neuraxial analgesia.
• Intrapartum ultrasound measurements included head-to-symphysis distance (HSD), degree of asynclitism (AD), angle of progression (AoP), and midline angle (MLA).
• The AIDA algorithm, employing machine learning (MLP, RF, SVM, XGBoost, LR, DT), classified patients and predicted labor outcomes (ICD vs. non-ICD).

Main Results:

• A degree of asynclitism greater than 70 mm was significantly associated with increased cesarean delivery rates.
• Weak to very weak correlations were observed between AD and AoP, HSD, and MLA.
• The AIDA algorithm demonstrated high accuracy in predicting labor outcomes, with specific ML algorithms showing excellent performance in certain classes (e.g., RF for class 3 at 92% accuracy).

Conclusions:

• Degree of asynclitism, combined with HSD, MLA, and AoP, is a significant predictor of labor dystocia and outcomes.
• The AIDA algorithm shows promise as a decision support tool for predicting labor outcomes, potentially aiding in the reduction of unnecessary cesarean deliveries.
• Further validation with larger cohorts is recommended to refine AD thresholds and AIDA algorithm parameters.