Introduction to Machine Learning in Obstetrics and Gynecology

Machine Learning: A Potential Therapeutic Tool to Facilitate Neonatal Therapeutic Decision Making

Bacterial infection is one of the major causes of neonatal morbidity and mortality worldwide. Finding rapid and reliable methods for early recognition and diagnosis of bacterial infections and early individualization of antibacterial drug administration are essential to eradicate these infections and prevent serious complications. However, this is often difficult to perform due to non-specific clinical presentations, low accuracy of current diagnostic methods, and limited knowledge of neonatal pharmacokinetics. Although neonatal medicine has been relatively late to embrace the benefits of machine learning (ML), there have been some initial applications of ML for the early prediction of neonatal sepsis and individualization of antibiotics. This article provides a brief introduction to ML and discusses the current state of the art in diagnosing and treating neonatal bacterial infections, gaps, potential uses of ML, and future directions to address the limitations of current studies. Neonatal bacterial infections involve a combination of physiologic development, disease expression, and treatment response outcomes. To address this complex relationship, future models could consider appropriate ML algorithms to capture time series features while integrating influences from the host, microbes, and drugs to optimize antimicrobial drug use in neonates. All models require prospective clinical trials to validate their clinical utility before clinical use.

Prediction of metabolic syndrome following a first pregnancy

BACKGROUND

The prevalence of metabolic syndrome is rapidly increasing in the United States. We hypothesized that prediction models using data obtained during pregnancy can accurately predict the future development of metabolic syndrome.

OBJECTIVE

This study aimed to develop machine learning models to predict the development of metabolic syndrome using factors ascertained in nulliparous pregnant individuals.

STUDY DESIGN

This was a secondary analysis of a prospective cohort study (Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-Be Heart Health Study [nuMoM2b-HHS]). Data were collected from October 2010 to October 2020, and analyzed from July 2023 to October 2023. Participants had in-person visits 2 to 7 years after their first delivery. The primary outcome was metabolic syndrome, defined by the National Cholesterol Education Program Adult Treatment Panel III criteria, which was measured within 2 to 7 years after delivery. A total of 127 variables that were obtained during pregnancy were evaluated. The data set was randomly split into a training set (70%) and a test set (30%). We developed a random forest model and a lasso regression model using variables obtained during pregnancy. We compared the area under the receiver operating characteristic curve for both models. Using the model with the better area under the receiver operating characteristic curve, we developed models that included fewer variables based on SHAP (SHapley Additive exPlanations) values and compared them with the original model. The final model chosen would have fewer variables and noninferior areas under the receiver operating characteristic curve.

RESULTS

A total of 4225 individuals met the inclusion criteria; the mean (standard deviation) age was 27.0 (5.6) years. Of these, 754 (17.8%) developed metabolic syndrome. The area under the receiver operating characteristic curve of the random forest model was 0.878 (95% confidence interval, 0.846-0.909), which was higher than the 0.850 of the lasso model (95% confidence interval, 0.811-0.888; P<.001). Therefore, random forest models using fewer variables were developed. The random forest model with the top 3 variables (high-density lipoprotein, insulin, and high-sensitivity C-reactive protein) was chosen as the final model because it had the area under the receiver operating characteristic curve of 0.867 (95% confidence interval, 0.839-0.895), which was not inferior to the original model (P=.08). The area under the receiver operating characteristic curve of the final model in the test set was 0.847 (95% confidence interval, 0.821-0.873). An online application of the final model was developed (https://kawakita.shinyapps.io/metabolic/).

CONCLUSION

We developed a model that can accurately predict the development of metabolic syndrome in 2 to 7 years after delivery.

Predicting cesarean delivery among gravidas with morbid obesity-a machine learning approach

BACKGROUND

Women with obesity have higher rates of complications following cesarean delivery, such as wound infection and endometritis, with risks being the highest if a cesarean delivery is performed after labor. Previous efforts at predicting whether a patient's labor course would ultimately result in cesarean delivery have been intermediate with area under the curve in the 0.75 to 78 range.

OBJECTIVE

This study aimed to assess whether machine learning algorithms would outperform traditional modeling in developing a cesarean delivery prediction model among gravidas with morbid obesity (body mass index of ≥40 kg/m2) to determine whether a primary cesarean delivery may be beneficial.

STUDY DESIGN

This was a secondary analysis of a retrospective cohort of 1298 patients with morbid obesity presenting for vaginal delivery at ≥37 weeks of gestation between 2011 and 2016 at a single institution. Data available at the time of admission and delivery were modeled using logistic regression, decision tree, random forest, and support vector modeling with evaluation of area under the curve, accuracy, sensitivity, and specificity.

RESULTS

Logistic regression demonstrated an area under the curve of 0.816 (95% confidence interval, 0.810-0.817), which was superior to machine learning models when evaluating data at the time of delivery (demographic data, initial cervical examinations, comorbidities, and obstetrical interventions) (P<.001). However, there was no significant difference between most machine learning models and logistic regression area under the curve of 0.799 (95% confidence interval, 0.795-0.804) when evaluating parameters available at the time of admission (demographic data, initial cervical examinations, and comorbidities). Race was noted to be a significant predictor in both models (P<.001).

CONCLUSION

Machine learning and traditional modeling techniques are likely equivalent concerning cesarean delivery prediction in this population. The models developed showed good discrimination and may be used to guide clinical decision-making concerning the optimal mode of delivery.

Patient-Specific Pessaries for Pelvic Organ Prolapse Using Three-Dimensional Printing: A Pilot Study

IMPORTANCE

Vaginal pessaries are an effective nonsurgical treatment for pelvic organ prolapse (POP) when properly fitted. However, pessary fitting and use are often unsuccessful or imperfect.

OBJECTIVE

The objective of this study was to assess the feasibility of using patient-specific pessaries fabricated from three-dimensional (3D)-printed molds to improve POP symptoms and increase overall satisfaction of pessary treatment in patients using standard vaginal pessaries.

STUDY DESIGN

Patients undergoing POP treatment with standard vaginal pessaries were enrolled in this pilot prospective study. Patient-specific pessaries were designed and fabricated for each patient using patient input, physician input, and anatomic measurements from clinical assessment. Pessary fabrication involved injection of biocompatible liquid silicone rubber into 3D-printed molds followed by a biocompatible silicone coating. Pelvic organ prolapse symptomatic distress and pessary treatment satisfaction were evaluated before and after a 3-week patient-specific pessary home trial using the validated Pelvic Organ Prolapse Distress Inventory-6 form and a visual analog scale, respectively.

RESULTS

Eight women were included in this study. Changing from standard pessary to patient-specific pessary treatment was associated with an improvement in prolapse symptoms on the Pelvic Organ Prolapse Distress Inventory-6 (median change, -3.5; interquartile range, -5 to -2.5; P = 0.02) and an increase in overall pessary satisfaction on a visual analog scale (median change, +2.0; interquartile range, +1.0 to +3.0; P = 0.02). All patients reported either an improvement or no change in pessary ease of use, comfort, and the feeling of support provided by the pessary.

CONCLUSION

Patient-specific vaginal pessaries are a promising alternative to standard pessaries for alleviating POP symptoms and improving patient satisfaction with pessary use.

Use of artificial intelligence and deep learning in fetal ultrasound imaging

Deep learning is considered the leading artificial intelligence tool in image analysis in general. Deep-learning algorithms excel at image recognition, which makes them valuable in medical imaging. Obstetric ultrasound has become the gold standard imaging modality for detection and diagnosis of fetal malformations. However, ultrasound relies heavily on the operator's experience, making it unreliable in inexperienced hands. Several studies have proposed the use of deep-learning models as a tool to support sonographers, in an attempt to overcome these problems inherent to ultrasound. Deep learning has many clinical applications in the field of fetal imaging, including identification of normal and abnormal fetal anatomy and measurement of fetal biometry. In this Review, we provide a comprehensive explanation of the fundamentals of deep learning in fetal imaging, with particular focus on its clinical applicability. © 2022 International Society of Ultrasound in Obstetrics and Gynecology.

Middle eastern college of obstetricians and gynecologists (MCOG) practice guidelines: Role of prediction models in management of trial of labor after cesarean section. Practice guideline no. 05-O-22✰,✰✰,★,★★

Cesarean delivery rates have been steadily rising since the beginning of the 21st century. The growing incidence is even more prominent in developing countries owing to lack of evidence-based guidance and audit, and the expansion of private practice. The uprise in Cesarean delivery rate has been associated with considerable financial burden and has increased the risk otherwise uncommon serious complications such as placenta accreta disorders and uterine rupture. In addition to primary prevention of Cesarean delivery, trial of labor after cesarean section is one of the most successful strategies to reduce Cesarean deliveries and minimize risks associated with higher order Cesarean deliveries. This guideline appraises patient selection strategies and use of prediction model to promote counseling and enhance safety in women considering vaginal birth after Cesarean.

Endometrial Cancer Individualized Scoring System (ECISS): A machine learning-based prediction model of endometrial cancer prognosis

OBJECTIVE

To establish a prognostic model for endometrial cancer (EC) that individualizes a risk and management plan per patient and disease characteristics.

METHODS

A multicenter retrospective study conducted in nine European gynecologic cancer centers. Women with confirmed EC between January 2008 to December 2015 were included. Demographics, disease characteristics, management, and follow-up information were collected. Cancer-specific survival (CSS) and disease-free survival (DFS) at 3 and 5 years comprise the primary outcomes of the study. Machine learning algorithms were applied to patient and disease characteristics. Model I: pretreatment model. Calculated probability was added to management variables (model II: treatment model), and the second calculated probability was added to perioperative and postoperative variables (model III).

RESULTS

Of 1150 women, 1144 were eligible for 3-year survival analysis and 860 for 5-year survival analysis. Model I, II, and III accuracies of prediction of 5-year CSS were 84.88%/85.47% (in train and test sets), 85.47%/84.88%, and 87.35%/86.05%, respectively. Model I predicted 3-year CSS at an accuracy of 91.34%/87.02%. Accuracies of models I, II, and III in predicting 5-year DFS were 74.63%/76.72%, 77.03%/76.72%, and 80.61%/77.78%, respectively.

CONCLUSION

The Endometrial Cancer Individualized Scoring System (ECISS) is a novel machine learning tool assessing patient-specific survival probability with high accuracy.

Applying machine learning methods to psychosocial screening data to improve identification of prenatal depression: Implications for clinical practice and research

We utilized machine learning (ML) methods on data from the PROMOTE, a novel psychosocial screening tool, to quantify risk for prenatal depression for individual patients and identify contributing factors that impart greater risk for depression. Random forest algorithms were used to predict likelihood for being at high risk for prenatal depression (Edinburgh Postnatal Depression Scale; EPDS ≥ 13 and/or positive self-injury item) using data from 1715 patients who completed the PROMOTE. Performance matrices were calculated to assess the ability of the PROMOTE to accurately classify patients. Probability for depression was calculated for individual patients. Finally, recursive feature elimination was used to evaluate the importance of each PROMOTE item in the classification of depression risk. PROMOTE data were successfully used to predict depression with acceptable performance matrices (accuracy = 0.80; sensitivity = 0.75; specificity = 0.81; positive predictive value = 0.79; negative predictive value = 0.97). Perceived stress, emotional problems, family support, age, major life events, partner support, unplanned pregnancy, current employment, lifetime abuse, and financial state were the most important PROMOTE items in the classification of depression risk. Results affirm the value of the PROMOTE as a psychosocial screening tool for prenatal depression and the benefit of using it in conjunction with ML methods. Using such methods can help detect underreported outcomes and identify what in patients' lives makes them more vulnerable, thus paving the way for effective individually tailored precision medicine.