One of the hardest processes to overcome during labor is the positioning of the fetal
head during engagement and progression in the birth canal. These malpositions and
malrotations are also common causes of dystocic labor and arrest of progression, which
necessitates a surgical delivery. Traditionally, a vaginal digital examination is used to
assess the fetal head position and engagement. The degree of fetal engagement and
progression is delineated by different planes of the pelvis, which have values ranging
from -5 (5 cm above the ischial spines) to +5 (5 cm below the ischial spines with the
fetal head visible at the introitus). The head is considered engaged when the leading
point of the skull touches the ischial spine plane; this is referred to as station 0.
Moreover, when the parietal bone is the presenting feature, asynclitism is identified.
When the front parietal bone manifests, the condition is identified as anterior
asynclitism; when the posterior parietal bone presents, the condition is described as
posterior asynclitism. The fetal head's minor "tilted" posture in the birth canal,
however, is a result of the fetal head's physiological adjustment to the mother's pelvis
throughout labor. Malposition and malrotation of the fetal head in the birth canal,
characterized by significant asynclitism, might result in dystocia that necessitates
surgical delivery. The 15% of people have asynclitism, with anterior asynclitism being
more common than posterior asynclitism. The traditional classification of degree of fetal
engagement and progression has been criticized as inaccurate and poorly reproducible.
Furthermore, the failure of instrumental delivery, which may reach up to 10%. Errors in
the diagnosis of the fetal head station can have major implications during labor and
adverse perinatal outcomes, such as acidemia, fetal traumas, intracranial hemorrhages,
and low Apgar scores after an emergency cesarean section for failed operative vaginal
deliveries. Many researchers have proposed many ultrasonographic parameters, collected
during labor and delivery, to evaluate the engagement, descent, and internal rotation of
the fetal head in the birth canal. Thus, we used intrapartum ultrasound parameters,
measured in all the women during labor and recorded to be measured by artificial
intelligence and machine learning algorithms, called AIDA (Artificial Intelligence
Dystocia Algorithm), which incorporates a human-in-the-loop approach, that is, to use AI
(artificial intelligence) algorithms that prioritize the physician's decision and
explainable artificial intelligence (XAI). The AIDA was structured into five classes.
After a number of "geometric parameters" were collected, the data obtained from the AIDA
analysis were entered into a red, yellow, or green zone, linked to the analysis of the
progress of labor. Using the AIDA analysis, we were able to identify five reference
classes for patients in labor, each of which had a certain sort of birth outcome. A 100%
cesarean birth prediction was made in two of these five classes. The use of artificial
intelligence, through the evaluation of certain obstetric parameters in specific
decision-making algorithms, allows physicians to systematically understand how the
results of the algorithms can be explained. This approach can be useful in evaluating the
progress of labor and predicting the labor outcome, including spontaneous, whether
operative VD (vaginal delivery) should be attempted, or if ICD (intrapartum cesarean
delivery) is preferable or necessary. In this investigation, we will seek, with a
critical eye, to evaluate the implications of some geometric parameters measured using an
intrapartum ultrasound trying to highlight all the possible complications and problems of
a dystocic and eutocic labor, with an evaluation of a big volume of women in pregnancy
and during labor.
Currently, the Artificial Intelligence Dystocia Algorithm (AIDA) represents a significant
advancement in the application of artificial intelligence to intrapartum care. Developed
through two key studies, AIDA 1 and AIDA 2, this innovative novel approach combines and
integrates multiple geometric parameters measured through intrapartum ultrasonography
with machine learning techniques and algorithms to assess labor progress and predict
delivery outcomes. The selection of algorithms, methodologies for performance evaluation,
and the metrics employed were meticulously delineated in the respective publications on
AIDA 1 and AIDA 2. These seminal papers provide comprehensive elucidations of the machine
learning techniques utilized, the rationale underpinning their selection, and the
rigorous performance metrics applied to assess their efficacy in predicting labor
outcomes. The high values obtained demonstrate the robust discriminative capability of
the algorithms chosen in distinguishing between different delivery outcomes, underscoring
the potential clinical utility of the AIDA approach in obstetric decision-making.
The AIDA Methodology A two-step methodology was applied to the data sample. The initial
step, the correlation analysis, employed Pearson's correlation coefficient to ascertain
that the four geometric parameters exhibited negligible or statistically insignificant
correlations, ensuring each parameter contributed unique information regarding labor
progression. The subsequent step, the machine learning algorithm selection, involved
applying diverse supervised machine learning algorithms to the four geometric parameters
in conjunction with physician-determined delivery outcomes. The predictive performance of
each algorithm was quantified to identify the most efficacious models.
The AIDA Classification System The classification system is predicated upon the
identification of cut-off values for each geometric parameter associated with intrapartum
cesarean delivery (ICD) and non-ICD outcomes. A decision tree algorithm was utilized to
establish these cut-offs: for each parameter, values strongly associated with non-ICD
outcomes were designated as "green", those highly correlated with ICD were classified as
"red", and in cases where the data sample revealed intermediate ranges of uncertainty, a
"yellow" designation was applied.
Having assigned a color to each individual value for the four geometric parameters for
every parturition, the AIDA classification system employs a structured approach to
categorizing each labor event into one of five distinct classes. This color-coded
stratification of the geometric parameters facilitates a nuanced assessment of labor
progression, enabling a more refined classification of each case. AIDA class 0 denotes
all four parameters being within the green zone, indicating a high probability of non-ICD
outcomes. AIDA class 1 indicates one parameter being in the red or yellow zone and three
being in the green zone. AIDA class 2 signifies two parameters being in the red or yellow
zone and two being in the green zone. AIDA class 3 represents three parameters being in
the red or yellow zone and one being in the green zone. AIDA class 4 denotes all four
parameters being within the red or yellow zone, suggesting a heightened likelihood of
ICD.
The AIDA's Prediction Performance The integration of the delivery predictions obtained
from the three best performing algorithms with the AIDA classification system yielded
significantly improved results. A particularly salient finding was the algorithm's high
predictive accuracy for delivery outcomes, notably in AIDA classes 0 and 4. In AIDA class
0, characterized by all geometric parameters being within the green zone, the consistent
prediction of non-ICD outcomes suggests its potential utility in identifying cases where
intervention may be safely deferred. Conversely, the accurate prediction of ICD in AIDA
class 4 cases could expedite decision-making for cesarean delivery, potentially
optimizing maternal and fetal out-comes by mitigating the duration of prolonged,
unproductive labor. The final methodological step entails employing the most effective
machine learning algorithms for predicting delivery outcomes based on the four geometric
parameters' values with consideration of the relevant AIDA class. This approach enables
clinicians to evaluate a pre-diction's clinical reliability.
The remarkable predictive accuracy, particularly at the extremes of the AIDA
classification spectrum, underscores the potential of this algorithmic approach to
enhance clinical decision-making in labor management.
Potential Advantages of the AIDA in Dystocic Labor Management Dystocic labor, also known
as dystocia, broadly defined as difficult or obstructed labor, encompasses a range of
conditions that impede the normal progression of labor and delivery as forms of difficult
labor characterized by abnormally slow progress.
This condition can arise due to inefficient uterine contractions, abnormal fetal
presentation, or other complications that impede the normal process of childbirth.
Dystocia can lead to obstructed labor, where despite strong uterine contractions, the
fetus cannot descend through the birth canal due to an insurmountable barrier, often
occurring at the pelvic brim.
Several types of dystocia have been recognized in obstetrics, each with characteristics
and management implications, and the classification may vary slightly depending on the
medical literature or clinical approach: geometric dystocia, mechanical dystocia, dynamic
dystocia, fetal dystocia, uterine dystocia, functional dystocia, soft tissue dystocia,
compound presentations, maternal exhaustion dystocia, and labor dystocia.
Through its simultaneous measurement of four geometric parameters-the angle of
progression, asynclitism degree, head-symphysis distance, and midline angle-the AIDA
offers a comprehensive view of the spatial relationships between the fetus and the
maternal pelvis, provides an objective assessment of the fetal position, and holds
potential as a comprehensive tool for evaluating, directly or indirectly, various types
of dystocia documented in the medical literature.
This approach provides a more detailed and accurate picture of the progress of labor or
labor obstruction that might be missed by traditional assessment methods. By quantifying
parameters like the degree of asynclitism, the AIDA may enable earlier detection of fetal
malpositions that could lead to dystocia.
The AIDA classification system offers a nuanced approach to risk assessment, helping
clinicians tailor their management strategies.
