Induction of labor is a frequently performed obstetric procedure that involves the artificial initiation of uterine contractions before spontaneous onset. Its use has been increasing globally, particularly in high-income countries, where it accounts for about 25% of births in the European Union, the United Kingdom, and the United States. Induction is typically recommended for medical reasons such as post-term pregnancy (≥41 weeks), gestational diabetes, hypertensive disorders of pregnancy, premature rupture of membranes or concerns about fetal growth.

Home / Publications / Publication

Home / Publications / Publication

Ecografia

PREDICTION OF DELIVERY MODE AFTER INDUCTION USING MACHINE LEARNING MODELS

May 02, 25

Publication type: Article Summary
Original title: Predicting vaginal delivery after labor induction using machine learning: Development of a multivariable prediction model
Article publication date: November 2024
Source: Acta Obstetricia et Gynecologica Scandinavica
Authors: Iolanda Ferreira, Joana Simões, João Correia & Ana Luísa Areia

What is the goal, target audience, and areas of digital health it addresses?
     The main goal of the study is to develop and validate a machine learning model capable of predict the likelihood of vaginal delivery after induction of labor. The target audience includes obstetricians, healthcare professionals involved in maternity care, and clinical researchers. It addresses key areas of digital health such as predictive analytics, Artificial Intelligence (AI) in obstetrics, interpretability of algorithmic models, and personalized medicine.

What is the context?
     Induction of labor is a frequently performed obstetric procedure that involves the artificial initiation of uterine contractions before spontaneous onset. Its use has been increasing globally, particularly in high-income countries, where it accounts for about 25% of births in the European Union, the United Kingdom, and the United States. Induction is typically recommended for medical reasons such as post-term pregnancy (≥41 weeks), gestational diabetes, hypertensive disorders of pregnancy, premature rupture of membranes or concerns about fetal growth.

     Although it is a valuable clinical tool, induction can result in unplanned cesarean sections, which are associated with higher costs — including longer hospital stays and greater demand for healthcare professionals — as well as greater health risks – such as bleeding, infections, slower recovery, and future pregnancy complications. Cesarean section rates have increased significantly in many countries, with Portugal reaching 37.1% in 2021. Several factors contribute to this trend, including advanced maternal age, obesity, the use of assisted reproductive technologies, and cultural preferences.

What are the current approaches?
     Predicting the mode of delivery after induction is a significant clinical challenge, influenced by numerous variables. Traditional predictive tools such as the Bishop score — which assesses the condition of the cervix based on dilation (opening of the cervix), percentage of effacement (shortening of the thickness of the cervix), position and consistency of the cervix, and level of fetal descent in the pelvic canal — offer a limited view of the labor progression, as they do not consider relevant maternal and obstetric factors, such as the number of previous births, interpregnancy intervals, and complications during pregnancy. Similarly, conventional statistical models have demonstrated limitations in their accuracy and data comprehensiveness, often based on small datasets, using time as the primary predictor, and do not distinguish between spontaneous and induced labor. As a result, they show low accuracy, are ineffective in reducing cesarean rates, and have limited clinical adherence, as healthcare professionals consider variables that these models overlook.

     To address these gaps, recent studies have explored machine learning approaches capable of analyzing large datasets and capturing complex, non-linear relationships. Although some existing models have shown good performance, they still face limitations such as bias, mixing data from both spontaneous and induced labor, and lacking rigorous validation (e.g., cross-validation or testing on independent datasets). Nevertheless, multifactorial machine learning models represent a promising step toward more accurate and personalized prediction tools.

What does innovation consist of? How is the impact of this study assessed?
     The innovation of this study was the development of a multivariable predictive model based on machine learning, specifically focused on predicting the mode of delivery after labor induction. In addition to predicting the outcome, the model also identifies which clinical features have the greatest influence on these predictions, promoting interpretability.

     To develop the model, a comprehensive database was used with the clinical data from 2,434 pregnant women at term (between 37 to 42 weeks of gestation) with a singleton pregnancy who underwent labor induction at the Obstetrics Department of University Hospital of Coimbra between 2018 and 2021. Data were collected at different times: during antenatal visits, hospital admission, and throughout labor. Initially they resulted in a comprehensive dataset of 65 variables. These variables were grouped into four categories: maternal factors (age, height, medical history), inter-pregnancy data (interval between pregnancies, type of delivery in previous pregnancies), current pregnancy details (weight, body mass index, complications), and induction specifics (gestational age, Bishop score, reason and method of induction).

     The evaluation of the impact and performance of this study was carried out by comparing the performance of several machine learning models: Logistic Regression, Random Forest, Multilayer Perceptron, Support Vector Machine, AdaBoost and XGBoost.

     To ensure the consistency, robustness, and generalizability of the models to data not observed during training, a 10-fold cross-validation was used: the data was split into 10 parts, using 9 to train the model and the remaining part to test its performance — repeating this process until each part had been used for testing once. To assess performance variability and increase confidence in the results, this full cycle of cross-validation was repeated 30 times with random initializations.

     The performance of each model was evaluated and compared using key metrics: AUROC (Area Under the Receiver Operating Characteristic Curve), which measures the model’s ability to distinguish between vaginal and cesarean deliveries — where 1.0 indicates perfect discrimination and 0.5 represents random chance; specificity, which indicates the proportion of actual vaginal deliveries the model correctly identified; and sensitivity, which indicates the proportion of actual cesarean cases the model correctly identified. To enhance clinical relevance and transparency, the study also used SHAP (SHapley Additive exPlanations), an Explainable AI technique, that interprets model predictions by explaining the outcome for each patient. SHAP quantifies how much each variable contributed to a specific prediction, while also accounting for interactions between variables.

What are the main results? What is the future of this approach?
     The study found that 1,736 women (71.3%) delivered vaginally after induction, while 698 (28.7%) required an unplanned cesarean. Women who underwent cesarean delivery were generally older, had a higher body mass index, were less likely to have had previous births, more likely to have had a previous cesarean section, and were less physically prepared for labor, as indicated by lower Bishop scores. The most frequent indications for induction were concerns about fetal growth and prolonged pregnancies (≥ 41 weeks). Cesareans sections were mainly performed due to fetal distress or failure to progress during labor.

     The best-performing predictive model was Logistic Regression, with an AUROC of 0.794, indicating good discrimination ability. It also showed high specificity, correctly identifying 91% of vaginal deliveries — providing reassurance to women with a low predicted risk of cesarean after induction. The sensitivity was 0.766, showing good ability to detect actual cesarean cases. The most important predictors of successful vaginal delivery, according to the SHAP analysis, were a higher Bishop score, history of previous full-term deliveries, taller maternal height, shorter time between pregnancies, and history of vaginal delivery. These features are largely consistent with established clinical knowledge but are now quantitatively confirmed by the model.

     In the future, the model should be externally validated in other hospitals using different datasets to confirm its generalizability before widespread adoption. It could be integrated into clinical practice as a decision support tool, providing personalized predictions of the mode of delivery after induction based on individual data. This approach could enhance counseling for pregnant women, promote shared decision-making between women and healthcare professionals, and help simulate outcomes at different gestational ages, guiding the optimal timing for induction. In the long term, this model may contribute to reducing unnecessary cesarean sections, optimizing hospital resources, and improving maternal and neonatal outcomes.

Would you like to know all the details?

Know more here

Do you have an innovative idea in healthcare field?

Share it with us and see it come to life.
We will help bring your projects to life!

Share my idea

Newsletter

Receive the latest updates from the InovarSaúde portal.

Subscribe

Support

FAQs

Contacts

User Manual

República Portuguesa logo
logotipo SNS
SPMS logotipo

Follow Us

YouTube
LinkedIn

Co-funded by

PRR Logotipo
república Portuguesa logo
União Europeia Logo

Newsletter

Receive the latest updates from the InovarSaúde portal.

Subscribe

Support

FAQs

Contacts

User Manual

República Portuguesa logo
SNS Logo
SPMS Logo

Follow Us

Co-funded by

PRR Logotipo
República Portuguesa logo
União Europeia Logo

Home / Publications / Publication

Ecografia

PREDICTION OF DELIVERY MODE AFTER INDUCTION USING MACHINE LEARNING MODELS

May 02, 25

Publication type: Article Summary
Original title: Predicting vaginal delivery after labor induction using machine learning: Development of a multivariable prediction model
Article publication date: November 2024
Source: Acta Obstetricia et Gynecologica Scandinavica
Authors: Iolanda Ferreira, Joana Simões, João Correia & Ana Luísa Areia

What is the goal, target audience, and areas of digital health it addresses?
     The main goal of the study is to develop and validate a machine learning model capable of predict the likelihood of vaginal delivery after induction of labor. The target audience includes obstetricians, healthcare professionals involved in maternity care, and clinical researchers. It addresses key areas of digital health such as predictive analytics, Artificial Intelligence (AI) in obstetrics, interpretability of algorithmic models, and personalized medicine.

What is the context?
     Induction of labor is a frequently performed obstetric procedure that involves the artificial initiation of uterine contractions before spontaneous onset. Its use has been increasing globally, particularly in high-income countries, where it accounts for about 25% of births in the European Union, the United Kingdom, and the United States. Induction is typically recommended for medical reasons such as post-term pregnancy (≥41 weeks), gestational diabetes, hypertensive disorders of pregnancy, premature rupture of membranes or concerns about fetal growth.

     Although it is a valuable clinical tool, induction can result in unplanned cesarean sections, which are associated with higher costs — including longer hospital stays and greater demand for healthcare professionals — as well as greater health risks – such as bleeding, infections, slower recovery, and future pregnancy complications. Cesarean section rates have increased significantly in many countries, with Portugal reaching 37.1% in 2021. Several factors contribute to this trend, including advanced maternal age, obesity, the use of assisted reproductive technologies, and cultural preferences.

What are the current approaches?
     Predicting the mode of delivery after induction is a significant clinical challenge, influenced by numerous variables. Traditional predictive tools such as the Bishop score — which assesses the condition of the cervix based on dilation (opening of the cervix), percentage of effacement (shortening of the thickness of the cervix), position and consistency of the cervix, and level of fetal descent in the pelvic canal — offer a limited view of the labor progression, as they do not consider relevant maternal and obstetric factors, such as the number of previous births, interpregnancy intervals, and complications during pregnancy. Similarly, conventional statistical models have demonstrated limitations in their accuracy and data comprehensiveness, often based on small datasets, using time as the primary predictor, and do not distinguish between spontaneous and induced labor. As a result, they show low accuracy, are ineffective in reducing cesarean rates, and have limited clinical adherence, as healthcare professionals consider variables that these models overlook.

     To address these gaps, recent studies have explored machine learning approaches capable of analyzing large datasets and capturing complex, non-linear relationships. Although some existing models have shown good performance, they still face limitations such as bias, mixing data from both spontaneous and induced labor, and lacking rigorous validation (e.g., cross-validation or testing on independent datasets). Nevertheless, multifactorial machine learning models represent a promising step toward more accurate and personalized prediction tools.

What does innovation consist of? How is the impact of this study assessed?
     The innovation of this study was the development of a multivariable predictive model based on machine learning, specifically focused on predicting the mode of delivery after labor induction. In addition to predicting the outcome, the model also identifies which clinical features have the greatest influence on these predictions, promoting interpretability.

     To develop the model, a comprehensive database was used with the clinical data from 2,434 pregnant women at term (between 37 to 42 weeks of gestation) with a singleton pregnancy who underwent labor induction at the Obstetrics Department of University Hospital of Coimbra between 2018 and 2021. Data were collected at different times: during antenatal visits, hospital admission, and throughout labor. Initially they resulted in a comprehensive dataset of 65 variables. These variables were grouped into four categories: maternal factors (age, height, medical history), inter-pregnancy data (interval between pregnancies, type of delivery in previous pregnancies), current pregnancy details (weight, body mass index, complications), and induction specifics (gestational age, Bishop score, reason and method of induction).

     The evaluation of the impact and performance of this study was carried out by comparing the performance of several machine learning models: Logistic Regression, Random Forest, Multilayer Perceptron, Support Vector Machine, AdaBoost and XGBoost.

     To ensure the consistency, robustness, and generalizability of the models to data not observed during training, a 10-fold cross-validation was used: the data was split into 10 parts, using 9 to train the model and the remaining part to test its performance — repeating this process until each part had been used for testing once. To assess performance variability and increase confidence in the results, this full cycle of cross-validation was repeated 30 times with random initializations.

     The performance of each model was evaluated and compared using key metrics: AUROC (Area Under the Receiver Operating Characteristic Curve), which measures the model’s ability to distinguish between vaginal and cesarean deliveries — where 1.0 indicates perfect discrimination and 0.5 represents random chance; specificity, which indicates the proportion of actual vaginal deliveries the model correctly identified; and sensitivity, which indicates the proportion of actual cesarean cases the model correctly identified. To enhance clinical relevance and transparency, the study also used SHAP (SHapley Additive exPlanations), an Explainable AI technique, that interprets model predictions by explaining the outcome for each patient. SHAP quantifies how much each variable contributed to a specific prediction, while also accounting for interactions between variables.

What are the main results? What is the future of this approach?
     The study found that 1,736 women (71.3%) delivered vaginally after induction, while 698 (28.7%) required an unplanned cesarean. Women who underwent cesarean delivery were generally older, had a higher body mass index, were less likely to have had previous births, more likely to have had a previous cesarean section, and were less physically prepared for labor, as indicated by lower Bishop scores. The most frequent indications for induction were concerns about fetal growth and prolonged pregnancies (≥ 41 weeks). Cesareans sections were mainly performed due to fetal distress or failure to progress during labor.

     The best-performing predictive model was Logistic Regression, with an AUROC of 0.794, indicating good discrimination ability. It also showed high specificity, correctly identifying 91% of vaginal deliveries — providing reassurance to women with a low predicted risk of cesarean after induction. The sensitivity was 0.766, showing good ability to detect actual cesarean cases. The most important predictors of successful vaginal delivery, according to the SHAP analysis, were a higher Bishop score, history of previous full-term deliveries, taller maternal height, shorter time between pregnancies, and history of vaginal delivery. These features are largely consistent with established clinical knowledge but are now quantitatively confirmed by the model.

     In the future, the model should be externally validated in other hospitals using different datasets to confirm its generalizability before widespread adoption. It could be integrated into clinical practice as a decision support tool, providing personalized predictions of the mode of delivery after induction based on individual data. This approach could enhance counseling for pregnant women, promote shared decision-making between women and healthcare professionals, and help simulate outcomes at different gestational ages, guiding the optimal timing for induction. In the long term, this model may contribute to reducing unnecessary cesarean sections, optimizing hospital resources, and improving maternal and neonatal outcomes.

Would you like to know all the details?

Know more here
Perturbação do sono

A Digital Intervention for Insomnia in Oncology

Insomnia is a sleep disorder characterised by persistent difficulties in initiating sleep, maintaining sleep during the night, or achieving restful sleep. These difficulties arise even in the presence of adequate sleeping conditions and are often…

Read more
Sistema de telemonitorização remota

Digital Technology Revolutionising Post-cardiac Surgery

According to the World Health Organisation, cardiovascular disease remains the leading cause of death worldwide, responsible for around 17.9 million deaths a year. Its high prevalence is associated with unhealthy lifestyles characterised by poor diet,…

Read more
Sistema robótico autónomo INSIDE

Autonomous Robotics System for Autism Therapy

Autism spectrum disorder is a neurodevelopmental condition with significant clinical, social and economic repercussions throughout life. According to the World Health Organization, it is estimated to affect approximately 1 in 160 children worldwide. Its origin…

Read more
Enfermeira com um telefone

Mobile Application to Improve Workflows in Nursing Homes

Portugal has one of the highest aging populations in the world, placing increasing pressure on elderly care services, especially in nursing homes. Healthcare professionals in these facilities are often overwhelmed due to the increasing number…

Read more
troca de informações de saúde e interoperabilidade

New Era of Interoperability in Healthcare Systems

The growing use of electronic health records, digital diagnostic systems and remote monitoring technologies has led to a significant increase in the volume and complexity of health data. This increase intensifies the need for continuous,…

Read more
robótica colaborativa

Collaborative Robotics Improves Working Conditions

Workers face growing challenges in the industrial environment. Among the most critical are fatigue and inappropriate postures, often associated with repetitive tasks and working conditions that lack ergonomic suitability. These factors represent significant risks for…

Read more
Benefícios da Eletrônica Médica

Detection of Anxiety and Panic Attacks in Real Time

The growing number of people with anxiety disorders, along with increased awareness of mental health, drives the need for new technological tools that provide remote and continuous monitoring of anxiety and panic disorders. Thus, the…

Read more
tele-ecografia

A Novel Approach for Robotic-assisted Tele-echography

Currently, robotic systems for ultrasound diagnostic procedures fall into two main categories: portable robots that require manual positioning and fully autonomous robotic systems that independently control the ultrasound probe’s orientation and positioning. Portable robots rely…

Read more
Personalização e tecnologia na gestão da Diabetes

Personalization and Technology in Diabetes Management

IPDM has significant potential to improve diabetes management and drive health system reforms to become high-performing, effective, equitable, accessible, and sustainable. Evidence and good practices inspire health system transformation. Adopting person-centred approaches like co-creation and…

Read more
TEF-HEALTH Logo

SPMS Integrates the TEF-Health Initiative

SPMS participates in the TEF-Health initiative as a partner in a consortium composed of 51 entities from 9 European Union countries. This action is co-financed by the European Commission and has a duration of five…

Read more
Global Digital Health Partnership Logo

SPMS Represents Portugal as Vice-president of GDHP

The GDHP is an intergovernmental organization in the digital health sector that facilitates cooperation and collaboration between government representatives and the World Health Organization (WHO). Its purpose is to foster policymaking that promote the digitalization…

Read more
Portugal INCoDe.2030

Digital Transformation of Health at INCoDe.2030 in Tomar

The “National Digital Skills Initiative e.2030, Portugal” (INCoDe.2030) is an initiative that aims to improve the Portuguese population’s level of digital skills, placing Portugal at the level of the most advanced European countries in this…

Read more
HealthData@PT Logo

HealthData@PT: New SPMS Initiative for Health Data

Action HealthData@PT is launched in the context of the implementation of the European Health Data Space, and is an initiative approved by the European Commission under the EU4Health 2021-2027 programme. This initiative contributes to the…

Read more

Do you have an innovative idea in healthcare field?

Share it with us and see it come to life.
We will help bring your projects to life!

Share my idea

Newsletter

Receive the latest updates from the Inovarsaúde portal.

Subscribe

Support

FAQs

Contacts

User Manual

República Portuguesa logo
SNS Logo
SPMS Logo

Follow Us

YouTube
LinkedIn

Co-funded by

PRR Logotipo
República Portuguesa logo
União Europeia Logo
Scroll to Top