Asynclitism, a misalignment of the fetal head with respect to the plane of passage through the birth canal, represents a significant obstetric challenge. High degrees of asynclitism are associated with labor dystocia, difficult operative delivery, and cesarean delivery. Despite its clinical relevance, the diagnosis of asynclitism and its influence on the outcome of labor remain matters of debate. This study analyzes the role of the degree of asynclitism (AD) in assessing labor progress and predicting labor outcome, focusing on its ability to predict intrapartum cesarean delivery (ICD) versus non-cesarean delivery. The study also aims to assess the performance of the AIDA (Artificial Intelligence Dystocia Algorithm) algorithm in integrating AD with other ultrasound parameters for predicting labor outcome. This retrospective study involved 135 full-term nulliparous patients with singleton fetuses in cephalic presentation undergoing neuraxial analgesia. Data were collected at three Italian hospitals between January 2014 and December 2020. In addition to routine digital vaginal examination, all patients underwent intrapartum ultrasound (IU) during protracted second stage of labor (greater than three hours). Four geometric parameters were measured using standard 3.5 MHz transabdominal ultrasound probes: head-to-symphysis distance (HSD), degree of asynclitism (AD), angle of progression (AoP), and midline angle (MLA). The AIDA algorithm, a machine learning-based decision support system, was used to classify patients into five classes (from 0 to 4) based on the values of the four geometric parameters and to predict labor outcome (ICD or non-ICD). Six machine learning algorithms were used: MLP (multi-layer perceptron), RF (random forest), SVM (support vector machine), XGBoost, LR (logistic regression), and DT (decision tree). Pearson's correlation was used to investigate the relationship between AD and the other parameters. A degree of asynclitism greater than 70 mm was found to be significantly associated with an increased rate of cesarean deliveries. Pearson's correlation analysis showed a weak to very weak correlation between AD and AoP (PC = 0.36, p < 0.001), AD and HSD (PC = 0.18, p < 0.05), and AD and MLA (PC = 0.14). The AIDA algorithm demonstrated high accuracy in predicting labor outcome, particularly for AIDA classes 0 and 4, with 100% agreement with physician-practiced labor outcome in two cases (RF and SVM algorithms) and slightly lower agreement with MLP. For AIDA class 3, the RF algorithm performed best, with an accuracy of 92%. AD, in combination with HSD, MLA, and AoP, plays a significant role in predicting labor dystocia and labor outcome. The AIDA algorithm, based on these four geometric parameters, has proven to be a promising decision support tool for predicting labor outcome and may help reduce the need for unnecessary cesarean deliveries, while improving maternal-fetal outcomes. Future studies with larger cohorts are needed to further validate these findings and refine the cut-off thresholds for AD and other parameters in the AIDA algorithm.

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

Cicinelli, Ettore;Vimercati, Antonella;D'Amato, Antonio;Baldini, Giorgio Maria;Dellino, Miriam;Trojano, Giuseppe;Difonzo, Tommaso;
2024-01-01

Abstract

Asynclitism, a misalignment of the fetal head with respect to the plane of passage through the birth canal, represents a significant obstetric challenge. High degrees of asynclitism are associated with labor dystocia, difficult operative delivery, and cesarean delivery. Despite its clinical relevance, the diagnosis of asynclitism and its influence on the outcome of labor remain matters of debate. This study analyzes the role of the degree of asynclitism (AD) in assessing labor progress and predicting labor outcome, focusing on its ability to predict intrapartum cesarean delivery (ICD) versus non-cesarean delivery. The study also aims to assess the performance of the AIDA (Artificial Intelligence Dystocia Algorithm) algorithm in integrating AD with other ultrasound parameters for predicting labor outcome. This retrospective study involved 135 full-term nulliparous patients with singleton fetuses in cephalic presentation undergoing neuraxial analgesia. Data were collected at three Italian hospitals between January 2014 and December 2020. In addition to routine digital vaginal examination, all patients underwent intrapartum ultrasound (IU) during protracted second stage of labor (greater than three hours). Four geometric parameters were measured using standard 3.5 MHz transabdominal ultrasound probes: head-to-symphysis distance (HSD), degree of asynclitism (AD), angle of progression (AoP), and midline angle (MLA). The AIDA algorithm, a machine learning-based decision support system, was used to classify patients into five classes (from 0 to 4) based on the values of the four geometric parameters and to predict labor outcome (ICD or non-ICD). Six machine learning algorithms were used: MLP (multi-layer perceptron), RF (random forest), SVM (support vector machine), XGBoost, LR (logistic regression), and DT (decision tree). Pearson's correlation was used to investigate the relationship between AD and the other parameters. A degree of asynclitism greater than 70 mm was found to be significantly associated with an increased rate of cesarean deliveries. Pearson's correlation analysis showed a weak to very weak correlation between AD and AoP (PC = 0.36, p < 0.001), AD and HSD (PC = 0.18, p < 0.05), and AD and MLA (PC = 0.14). The AIDA algorithm demonstrated high accuracy in predicting labor outcome, particularly for AIDA classes 0 and 4, with 100% agreement with physician-practiced labor outcome in two cases (RF and SVM algorithms) and slightly lower agreement with MLP. For AIDA class 3, the RF algorithm performed best, with an accuracy of 92%. AD, in combination with HSD, MLA, and AoP, plays a significant role in predicting labor dystocia and labor outcome. The AIDA algorithm, based on these four geometric parameters, has proven to be a promising decision support tool for predicting labor outcome and may help reduce the need for unnecessary cesarean deliveries, while improving maternal-fetal outcomes. Future studies with larger cohorts are needed to further validate these findings and refine the cut-off thresholds for AD and other parameters in the AIDA algorithm.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11586/507701
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? 0
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
social impact