Skip to content
Frontiers
Search

Review

Frontiers in Anesthesiology ·

Volume 5 - 2026 | https://doi.org/10.3389/fanes.2026.1896595

Artificial intelligence and obstetric anesthesia: a scoping review

  • Maeve F. Givens1
  • Edward Rodriguez Caceres1
  • Rohan Jotwani2
  • John E. Rubin2
  • Peter A. Goldstein2,3,4
  • Robert S. White2*corresponding author
  1. 1M.D. Program, Weill Cornell Medicine, Cornell University, New York, NY, United States
  2. 2Department of Anesthesiology, Weill Cornell Medicine, Cornell University, New York, NY, United States
  3. 3Department of Medicine, Weill Cornell Medicine, Cornell University, New York, NY, United States
  4. 4Feil Family Brain and Mind Research Institute, Weill Cornell Medicine, Cornell University, New York, NY, United States
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Artificial intelligence (AI) is expanding rapidly across anesthesiology, with growing applications in obstetric anesthesia at a time when maternal mortality in the United States remains high and marked racial and ethnic disparities persist. This scoping review aimed to characterize the current uses, emerging applications, and future directions of AI across the field of obstetric anesthesiology. We searched PubMed and Embase, combining clinical terms for obstetric anesthesia and delivery with methodological terms for AI and machine learning (ML). English language original studies, systematic reviews, clinical trials, and meta-analyses published between January 2015 and December 2025 describing the development, validation, or clinical application of AI/ML tools relevant to obstetric anesthesia were eligible. Records were screened by two reviewers with consensus adjudication by two board-certified anesthesiologists. Included studies were categorized by AI domain, ML class, intended application, validation approach, and aspect of care. Of 3,135 records identified, 172 met inclusion criteria for review, comprising 154 original research studies, 17 systematic or narrative reviews, and 1 case report. 12 other records encompassing 9 registered clinical trials or protocols and 3 workforce-focused studies were also identified and reported upon separately. 69% of publications appeared after 2022. The most common domains were maternal risk and complications prediction (n = 61), delivery mode prediction (n = 43), and patient education and conversational AI (n = 15). Reported model discrimination was frequently strong, with areas under the curve (AUC) commonly between 0.75 and 0.95. However, only 14% of the 154 original research studies reported external validation. Prospective evaluations were rare, and almost none described integration into clinical workflow. The use of AI in obstetric anesthesia has accelerated rapidly, spanning maternal risk prediction, delivery planning, fetal monitoring, patient education, and provider training. Although discrimination metrics are generally strong, the literature remains largely retrospective, with limited external validation and little evidence of real-world clinical impact. Future efforts must prioritize externally validated, prospectively evaluated, and workflow-integrated studies if AI is to meaningfully reduce preventable maternal morbidity and mortality.

Keywords

1 Introduction

Artificial intelligence (AI) is broadly considered as computer systems that perform cognitive activities such as learning and problem-solving (1). Subsets of AI include, but are not limited to, machine learning, natural language processing, and computer vision (2). Machine learning (ML) is defined broadly as when programs analyze, learn, and respond to data with three major approaches that can be classified as supervised, unsupervised, or reinforcement based learning (1–4). Numerous techniques exist within ML including decision tree learning, which uses a flowchart-based approach; neural networks, which mimic the brain by analyzing data through an interconnected layered structure; and, Bayesian methods, which use a statistical approach based on probability theory (1, 2, 4).

AI's use in anesthesiology is rapidly expanding, with applications spanning the perioperative landscape. AI can be deployed broadly to analyze electronic medical record data and radiographic images, provide predictive risk assessments, and aid in real-time clinical decision making (2, 5–8). In obstetrics and obstetrical anesthesiology, AI can be leveraged to reduce maternal morbidity and mortality, determine the risk of severe complications such as postpartum hemorrhage and placenta accreta spectrum, assist in care delivery by predicting severe hypotension, and evaluate optimal modes of delivery (9–12).

These applications are especially important as maternal mortality rates remain high and continue to increase in the United States with 32.6 deaths per 100,000 live births in 2022 (13). Significant racial and ethnic disparities in maternal outcomes and anesthesia approaches also persist (14–16). Deploying AI tools across the spectrum of obstetric anesthesiology offers critical opportunities to improve the quality of care and reduce pregnancy-related complications and mortality.

The use of AI in the perioperative landscape is rapidly expanding and evolving. However, obstetric anesthesiology poses unique questions in AI's deployment given the distinct challenges of the peripartum setting (i.e., dynamic nature of two-patient physiology, acute intrapartum shifts, and low baseline incidence of severe events like placenta accreta or cardiac arrest). This scoping review aims to characterize the applications of AI across the field of obstetric anesthesia to provide a guiding resource on its current uses, emerging applications, and future directions.

2 Methods

A literature search was conducted in PubMed and Embase. The search strategy centered on PubMed indexed papers and then used Embase for verification and to ensure adequate coverage of the relevant literature. The search combined two concept blocks with the Boolean operator “AND”: (i) a clinical block capturing obstetric anesthesia and delivery, using the relevant Medical Subject Headings (MeSH) and Emtree terms together with free-text language (e.g., “obstetric anaesthesia,” “labor analgesia,” “cesarean section,” “parturient”); and (ii) a methodological block capturing AI and ML, also combining relevant MeSH and Emtree terms with free-text terms (e.g., “deep learning,” “neural network,” “natural language processing,” “large language model,” “random forest,” “support vector machine,” “gradient boosting,” “predictive model,” “risk prediction”). An animal exclusion filter was applied to remove records indexed as non-human studies. No date or language restrictions were embedded in the search strings, as these were applied as eligibility criteria during screening. The complete search strings for both databases are provided in Supplementary Table 1.

Eligible records were reviewed, which included English-language articles published between January 2015 and December 2025. Articles included original studies, systematic reviews, and meta-analyses describing the development, validation, or clinical application of AI/ML tools relevant to obstetric anesthesia. Studies were excluded if they were not available in English, were conference abstracts, were grey literature, were pre-prints, could not have their full texts accessed, addressed AI outside of the maternal or obstetric anesthesia context, or did not include any mention of AI or ML methodology. Pre-prints, grey literature, or conference abstracts were not systematically searched or excluded prior to screening, but were excluded during the screening process itself. Clinical trials and original studies focused on the anesthesia workforce were reported upon separately.

Records were screened by two reviewers, first by title and abstract, and then by full text. Eligibility decisions were resolved by consensus by two board-certified anesthesiologists (PG and RW), including one fellowship-trained obstetric anesthesiologist (RW). Data from each included study that was analyzed included bibliographic details (identifiers, year of publication, source database, and publication type), the country of origin of research, the targeted AI domain, the class of ML (e.g., supervised classification), the specific techniques and algorithm categories employed, the intended application of the model, and the relevant aspect of obstetric anesthetic care. Data extraction and classification questions were settled via detailed review and consensus (MG and ER), with input from a board-certified anesthesiologist as necessary (RW). For each included study, the best-performing model and its principal performance metric (e.g., area under the curve) with the associated value and confidence interval, the validation approach (internal versus external), and the sample size were also recorded. Included studies were appraised qualitatively and categorized into the following AI domains: maternal risk and complications prediction, delivery mode prediction, fetal heart rate tracing analysis, imaging and radiomics, electronic health record and clinical documentation, patient education and conversational AI, and provider education and simulation. Retrospective characterization was used to create these domains after initial abstract review and prior to full text review. Table 1 provides a representative collection of articles by domain.

Table 1. Representative articles by AI domain.

AI DomainMain TopicAI ApplicationJournalAuthorYearReference NumberMain Result
Maternal Risks and Complications PredictionIntrapartum fever prediction using pre-labor analgesia clinical dataPrediction ModelBMC Pregnancy and ChildbirthLiu et al.2025(48)Multilayer perceptron model performed best (AUC of 0.71) at predicting intrapartum fever prior to administration of labor analgesia
Predicting the likelihood of developing significant hypotension and vasopressor requirement under spinal anesthesia when undergoing a Cesarean sectionPrediction ModelBMC AnesthesiologyGratz et al.2020(50)Neural Network approach performed best (AUC of 0.89) at predicting hypotension and vasopressor requirement
Postoperative pain prediction after spinal morphine administration using machine learningPrediction ModelHeliyonTan et al.2024(118)Ridge Regression model performed best (AUC of 0.72) at predicting postoperative pain
Patient Education/AI ChatbotAI chatbots vs traditional resources for labor epidural patient educationEducationalInternational Journal of Obstetric AnesthesiaGondode et al.2025(91)AI chatbots performed comparably or better than traditional resources on accuracy and completeness, but readability needs improvement
Assessment of LLMs for multilingual translating in obstetric anesthesia scenariosClinical Support ToolInternational Journal of Obstetric AnesthesiaLomas et al.2024(94)LLM translations showed reasonable accuracy but variable performance across languages; some clinically relevant errors identified
Fetal Heart Rate TracingUsing ML models to predict changes in fetal heart rate after neuraxial analgesia.Prediction ModelBMC Pregnancy and ChildbirthRiveros-Perez et al.2023(91)Random forest (RME of 0.92) models found use of CSE, presence of decelerations, total dose of bupivacaine, and total dose of vasopressors after CSE are associated with decreases in fetal heart rate in healthy parturients during labor
Predicting acidemia from fetal heart rate tracing data using ML techniquesPrediction ModelEntropy (Basel)Esteban-Escaño et al.2022(77)Random forest model performed best at predicting acidemia from FHR tracing data (AUROC of 0.87)
Using neural network models to predict fetal state from cardiotocography dataPrediction ModelArtificial Intelligence in MedicineIraji2019(119)DSSAE model performed best (accuracy of 99.5%) at predicting fetal state from cardiotocography data
Imaging and RadiomicsLumbar level identification in obstetric patients using novel ultrasound softwareClinical support toolCanadian Journal of AnesthesiaHetherington et al.2022(12)Novel software had higher agreement with traditional ultrasound than manual palpation did in identifying L3-4 and all other lumber interspaces after adjusting for BMI in healthy term obstetric patients
Using deep learning radiomics to diagnose placenta accreta with MRI imagingPrediction Model/Data ExtractionJournal of Magnetic Resonance ImagingZheng et al.2024(10)Combined DL/radiomics/clinical model (AUC of 0.85) improved PAS diagnosis and subtype identification compared with clinical or radiomics alone
Using deep learning and MRI radiomics to automatically segment cervical areas and predict intraoperative blood loss level during cesarean sectionsPrediction Model/Data ExtractionFrontiers in Bioengineering and BiotechnologyLiu et al.2020(120)Combined radiomics/deep learning model (accuracy of 75.6%) was able to predict level of bleeding in patients with pernicious placenta previa undergoing cesarean section
Provider Education and SimulationPerformance of LLM clinical decision support for high-risk anesthesia (obstetric and geriatric) casesTrainingJournal of Medical SystemsRuan et al.2025(98)LLMs (DeepSeek performed best) show promise as decision support tools for complex anesthesia care including obstetric cases, though performance varies; further research is needed before clinical integration
Review article focused on advances in training for neonatal resuscitation for obstetric anesthesiologistsTrainingCurrent Opinion in AnesthesiologyFantin et al.2025(99)Encourages obstetric anesthesiologists to adapt new training tools including AI skill retention tools
Delivery Mode PredictionPredicting mode of delivery in prolonged pregnancies using machine learningPrediction ModelScientific ReportsSchmauder et al.2025(56)Linear regression ML model performed best (AUC of 0.69) at predicting mode of delivery in prolonged pregnancies. Major predictors were maternal characteristics, parity, and gestational age
Predicting unplanned cesarean delivery using machine learningPrediction ModelInternational Journal of Gynecology and ObstetricsMeyer et al.2023(121)XGBoost ML model performed best (AUC of 0.84) at predicting unplanned cesarean deliveries
Predicting mode of delivery using deep neural network modellingPrediction ModelEuropean Journal of Obstetrics and Gynecology and Reproductive BiologyKuanar et al.2024(122)Deep neural network model successfully predicted (AUC of 0.99) mode of delivery. Major predictors included cervical length and fetal heart rate

A completed Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension Scoping Reviews (PRISMA-ScR) Checklist is included as a Supplementary Figure.

3 Results

3.1 Search results

The database searches identified 3,135 records (PubMed n = 1,850; Embase n = 1,285). After removal of 541 duplicates, 2,594 unique records were screened by title and abstract, of which 2,357 were excluded as off-topic. The remaining 237 records were assessed by full-text review, and 53 were excluded for the following reasons: conference abstracts (n = 36), letters to editors/editorials (n = 8), pre-prints (n = 2), non-English language (n = 4), and articles for which the full text could not be obtained (n = 3). Twelve articles (i.e., 9 clinical trials and 3 anesthesia workforce studies) were separated and reported discretely as described below. A total of 172studies met the inclusion criteria and were included in the scoping review. The full study selection process is summarized in Figure 1.

Flowchart of the methodology used to screen articles. 172 articles published between 2015−2025 were included in the review.
Figure 1. Flowchart of the methodology used to screen articles. 172 articles published between 2015−2025 were included in the review.

3.2 Study selection and overview

The 172 included studies comprised 154 original research studies, 17 narrative or systematic reviews, and 1 case report reporting on AI applications relevant to obstetric anesthesia. Another 12 records were reviewed and reported upon separately given their distinct scope: nine registered clinical trials or protocol descriptions addressing AI-based interventions in the obstetric or peripartum setting, and three studies addressing AI applications focused specifically on the anesthesiologist workforce (e.g., burnout prediction, residency personal-statement generation). These 12 articles were included in this review for descriptive purposes and were not used for calculating results such as article AI domain or validation status. They are presented separately here to inform the reader of advances in clinical trials and workforce-focused uses of AI in obstetric anesthesia.

3.3 Temporal and geographic distribution

Included studies were published between 2015 and 2025, with a marked acceleration in recent years. Roughly three-quarters (n = 120, 69%) appeared in 2023 or later, and more than half (n = 94, 54%) were published in 2024 or 2025 alone. Research was geographically diverse, with the United States (n = 40) and China (n = 35) contributing the largest single-country shares, followed by Israel (n = 9), Italy (n = 8), India (n = 5), Singapore (n = 5), and Spain (n = 5). Eighteen additional studies were multinational collaborations, and the remainder originated from a range of other countries.

3.4 Artificial intelligence approaches and validation

Among studies that reported a specific ML class, classification tasks were most common (n = 45 used classification alone, with an additional 44 combining classification with regression or neural-network approaches), followed by regression (n = 15 alone; frequently combined with other classes), neural networks (n = 13 alone), and large language models (n = 17). Unsupervised learning (n = 1), reinforcement learning (n = 1), and dedicated radiomics pipelines (n = 4, plus 1 combined with classification) were used infrequently.

Model validation was reported in most studies. Of the 154 original research studies, 117 (76%) reported some form of validation, 21 (13.6%) reported no validation, and 17 (11%) involved analyses for which validation was not applicable (chatbot-based applications, or qualitative analyses). The 17 narrative or systematic reviews and 1 case report did not report validation because it was not applicable and were not included in validation status calculations. Among studies that did report model validation, internal validation was most common: 46 used a held-out test or validation set, and 45 used cross-validation, with 4 reporting both. Only 22 studies (14% of all 154 original research studies) reported external validation in an independent dataset. The distribution of validation approaches by type of AI application is shown in Figure 2; this stratification highlights that validation rates were highest for prediction models, of which 92 of 107 (86%) reported some form of validation, but only 17 (16%) reported external validation, whereas clinical support tool applications more often reported no validation.

The distribution of validation approaches by type of AI application in the 154 original research articles.
Figure 2. The distribution of validation approaches by type of AI application in the 154 original research articles.

3.5 Distribution across AI domains and aspect of care

Studies were classified into one of eight categories: six single AI domains, a “multiple” category for studies addressing more than one domain, and an “other” category for studies that did not fall within these domains. The most frequently represented domain was maternal risk and complications prediction (n = 61), followed by delivery mode prediction (n = 43), patient education and conversational AI (n = 15), fetal heart rate tracing analysis (n = 12), imaging and radiomics (n = 8), and provider education and simulation (n = 2). Several studies spanned multiple domains (n = 14) and are included in the totals in each domain in the discussion. A further 17 studies addressed applications that did not fall within these categories and were grouped as “other.” The distribution of AI domains across publication years is shown in Figure 3, illustrating both the recent expansion of the field overall and the rapid emergence of patient-education and conversational-AI applications in 2024–2025.

The distribution of the 154 original research publications by obstetric anesthesia domain from 2015−2025.
Figure 3. The distribution of the 154 original research publications by obstetric anesthesia domain from 2015−2025.

When classified by aspects of care, the intrapartum period accounted for the largest share (n = 100), followed by studies spanning multiple phases (n = 43), the antepartum period (n = 12), and the postpartum period (n = 9); 10 studies addressed applications for which a single phase of care was not applicable. Trends in aspects of care by year of publication are shown in Figure 4.

The distribution of the 154 original research publications by aspect of care from 2015−2025.
Figure 4. The distribution of the 154 original research publications by aspect of care from 2015−2025.

3.6 Clinical trials and anesthesiologist-focused studies

The nine clinical trials identified included both completed studies and protocols, addressing AI applications in labor analgesia and its sequelae, lung ultrasound interpretation, automated spinal landmark identification, fetal heart rate analysis during neuraxial analgesia, and individualized risk scoring for placenta previa and delivery outcomes. The three anesthesiologist-focused studies addressed the use of neural networks to predict burnout among obstetric anesthesiologists and the use of generative AI in producing and evaluating obstetric anesthesiology fellowship personal statements.

4 Discussion

This review describes the scope of AI's use in the field of obstetric anesthesia by highlighting its current uses, emerging applications, and future directions. A wide breadth of AI applications was encountered in the 172 articles reviewed (154 original research studies, 17 narrative or systematic reviews, 1 case report). ML models and more broadly, AI tools spanned across obstetric anesthesia domains including maternal risk and complications, delivery mode, patient education and conversational AI, fetal heart rate tracing analysis, imaging and radiomics, and provider education and simulation. Below, we analyze the utilization of AI by domain, highlight existing challenges and future directions for AI, and discuss strengths and limitations of our study. Note, percentages are based upon the proportion of the 154 original research studies, 17 systematic or narrative reviews, and 1 case report that relate to each specific domain; the total does not include the 12 records (i.e., 9 clinical trials, 3 anesthesiology workforce studies) reported on separately at the end of the discussion

4.1 Maternal risk and complications

Nearly 40% of articles focused on maternal risks and complications, with 61 articles solely focused on this domain, and another 7 addressing both this domain and other domains including maternal risk and complications. Over the last decade, AI has been applied across the peripartum spectrum, including surgical site infections, uterine dehiscence, preterm premature rupture of membranes, and labor dystocia. ML is increasingly concentrated in several key areas: postpartum hemorrhage, placenta accreta spectrum, hypertensive disorders of pregnancy, cardiovascular complications, and neuraxial anesthesia complications.

4.1.1 Postpartum hemorrhage

Postpartum hemorrhage accounts for 11% of U.S. maternal mortality and is a leading preventable cause of maternal death and a key challenge for obstetric anesthesiologists; preparedness and accurate risk assessment are essential to reducing related morbidity and mortality (17, 18). Multiple supervised ML models have been developed to predict postpartum hemorrhage and transfusion, with widely varying performance (AUC 0.67 to 0.99) (19–21). Models also identified key risk factors that could enable earlier detection of high-risk patients (21). Large language models have been used to extract risk factors from prenatal notes and electronic health records (e.g., multiple gestation, prior cesarean) and to classify causes of hemorrhage (22, 23).

Important limitations remain. A review highlighted wide variation in hemorrhage rates across cohorts and models limited to specific delivery modes or conditions (24–26). Differing hemorrhage definitions across cohorts can markedly degrade performance on new populations: one model's C-statistic fell from 0.93 to 0.57 when applied to a contemporary cohort rather than an early-2000s cohort (27). There was also significant intra-study variation in model performance and variables reported. For example, one study found that using a linear regression ML model with variables such as fetal macrosomia, general anesthesia, and anemia best predicted PPH with an AUROC of 0.78, while an XGBoost classification model using variables such as living residency, parity, and induced labor predicted PPH with a AUROC of 0.98 (20). Changing definitions are a critical barrier to broadly applicable models (17, 18, 24). Although AI shows clear promise for hemorrhage prediction, greater external validation and training on large, diverse cohorts are still needed.

4.1.2 Placenta accreta spectrum

Placenta accreta spectrum is a life-threatening condition (28). Although its prevalence is low (∼0.17%), it drives severe peripartum morbidity, with pooled incidences of 52.2% for hysterectomy and 46.9% for transfusion (29). Antenatal identification is critical to optimizing outcomes (28). Radiomics and ML are promising for earlier recognition and risk assessment of severe complications (30–33). Several studies applied radiomics and MRI-based deep learning to assess placenta accreta spectrum (30, 31, 34). One 2025 study used a deep learning model using 15 radiomic features to predict placenta accreta spectrum reaching an AUC of 0.95. However, other machine learning models in that same study such as random forest or k-nearest neighbor using the same radiomic variables only achieved AUC values of 0.704 to 0.863 (30). Models predicting severe hemorrhage have achieved AUCs of 0.71 to 0.84 (33, 35).

4.1.3 Hypertensive disorders of pregnancy

Pre-eclampsia affects 555 per 10,000 deliveries, and hypertensive disorders account for 6.8% of pregnancy-related deaths (36). Assessing pre-eclampsia risk and severity is therefore important. A 2024 review of ML prediction found variable performance across 11 studies (AUCs 0.76 to 0.97) (37). A 2024 pilot study used routine biomarkers and clinical characteristics to predict severe preeclampsia, defined as preeclampsia, eclampsia, or Hemolysis, Elevated Liver Enzymes, Low Platelets (HELLP) syndrome requiring ICU treatment, and achieved an AUC of 0.91 using an XGBoost model (38). Aspartate aminotransferase, uric acid, and body mass index (BMI) were the strongest indicators of severe disease (38). Despite the small 81-patient cohort, the study illustrates ML's potential to detect patterns in complex data and support pre-eclampsia care planning (38).

4.1.4 Cardiomyopathies

Peripartum cardiomyopathy, a heart failure emerging during pregnancy or postpartum, is another leading cause of maternal mortality, complicating roughly 1 in 2000 pregnancies worldwide; affected patients have 76-fold higher odds of cardiogenic shock across the peripartum period (39, 40). AI has been used to identify peripartum cardiomyopathy and its complications. A recent trial found an AI-guided digital stethoscope doubled diagnosis of pregnancy-related cardiomyopathy versus usual care, and the number needed to screen to diagnose one new case of perinatal left ventricular systolic dysfunction was 47 (41, 42). More broadly, one ML model showed moderate discrimination for acute cardiovascular and renal complications in a large national dataset (AUC 0.68) (43).

4.1.5 Epidural-related maternal fever

Epidural-related maternal fever affects 15% to 25% of patients receiving neuraxial anesthesia and is difficult to distinguish clinically from other causes of intrapartum fever such as chorioamnionitis (44). Because of this diagnostic difficulty, it quadruples the risk of intrapartum conversion to cesarean delivery compared with those without fever (45). Multiple supervised ML models have been proposed to predict intrapartum fever, with variable success (AUCs 0.71 to 0.94) (46–49). One study found that adding inflammatory markers [neutrophil-lymphocyte ratio and white blood cell (WBC) count] improved risk assessment (46). Intra-study ML model performance variation was reported in this study using the same variables for each model. For example, an XGBoost model had an AUC of 0.82 while a decision tree model achieved an AUC of 0.92 in the testing dataset (46). These models are largely retrospective and not externally validated, limiting intrapartum utility, though the proposed frameworks could enable earlier, more targeted intervention if adapted clinically.

4.1.6 Complications of neuraxial anesthesia

ML has also been applied to other neuraxial complications such as post-neuraxial hypotension and breakthrough pain. A 2020 pilot neural network predicted significant post-spinal hypotension during cesarean section (AUC 0.89), comparable to an analytical approach based on arterial stiffness alone (AUC 0.87) (50). For breakthrough pain after neuraxial anesthesia, two ML models performed comparably to a multivariable regression model (all AUCs 0.763 to 0.772) (51). Larger datasets and further research are needed before such models outperform traditional statistical approaches.

4.2 Delivery mode

Approximately 30% of articles addressed prediction of delivery mode and timing with 43 articles individually falling within this domain and 9 addressing multiple domains including delivery mode. Studies spanned diverse contexts and countries and generally showed compelling predictive power (52). Models addressed antepartum and intrapartum prediction, including emergency cesarean, successful trial of labor after cesarean, and vaginal versus cesarean delivery in women without prior cesarean (53–60). Models also predicted onset and duration of labor and pre-term labor (61).

About one in three U.S. births occur by cesarean delivery (62). Most cesareans are unplanned; one study estimated they accounted for a quarter of all deliveries (63). Unplanned cesareans carry higher maternal and fetal complication and mortality rates than planned cesareans (64), with severe surgical morbidity more common after unplanned intrapartum than pre-labor planned cesareans (65). Multiple large-dataset studies evaluated ML for intrapartum delivery mode prediction, with variable performance (AUCs 0.69 to 0.932) (9, 58, 60, 66). In one retrospective study, the best model (K-nearest neighbor) predicted vaginal delivery four hours after admission with an AUC of 0.812; outperforming traditional logistic regression (60). Of note, other ML models such as random forest and decision tree models using different variable ratios performed worse with AUCs ranging from 0.581 to 0.756 (60). Despite ongoing accuracy concerns, ML offers personalized risk assessment for laboring patients, decision msupport, and better preparation of labor and delivery teams (58, 67).

Given the high rate of unplanned cesarean, predicting successful trial of labor after cesarean (TOLAC) is highly relevant to delivery planning and antepartum counseling (68). Accurate assessment matters because successful vaginal birth after cesarean (VBAC) lowers rates of hemorrhage, thromboembolism, and infection and speeds recovery, whereas a failed trial of labor causes greater morbidity than either successful VBAC or planned cesarean (68).

The ML models reviewed showed strong potential relative to older TOLAC calculators (55, 66). In one study, three ML models significantly outperformed the Maternal Fetal Medicine Unit (MFMU) model (66). A categorical boosted model predicted successful VBAC with AUCs of 0.767 (internal) and 0.707 (external validation) (55). This was comparable to logistic regression (AUC 0.762) and the updated 2021 MFMU calculator (AUC 0.75) (55, 69). Given these head-to-head results, further work should train and externally validate models on larger, more diverse cohorts (55, 66, 69). With such advances, ML models will likely surpass traditional TOLAC calculators and support more personalized risk stratification (9).

Labor dystocia is the most common indication for cesarean delivery globally (62). AI could identify cephalopelvic disproportion and dystocia earlier and improve management (70). A 2020 ML model outperformed a simple combination of maternal diabetes and estimated fetal weight for predicting dystocia (AUC 0.866 versus 0.784) (71). More recently, the Artificial Intelligence Dystocia Algorithm (AIDA) was developed to assess transverse fetal head position and associated cesarean risk (72). Although evaluated on only 135 cases, its random forest algorithm reached 95.5% accuracy (72). Collectively, these studies suggest ML will become a useful antepartum clinical tool. However, it is important to note that algorithm overreliance may create the clinical risk of unnecessary cesarean deliveries.

4.3 Fetal heart rate tracing analysis

Sixteen articles (9%) addressed fetal heart rate tracings, a mainstay of intrapartum monitoring that accounts for 27% of primary cesarean deliveries in the United States (73). 12 articles fell within this domain, and 4 addressed multiple domains including fetal heart tracing. The high false-positive rate and interobserver variability of cardiotocography (CTG) contribute to these rates (73–75). AI could offer real-time decision support and risk assessment for interpreting these tracings (74).

ML and deep learning models have increasingly been applied to CTG monitoring and to predict outcomes such as hypoxic-ischemic encephalopathy, fetal acidemia, and delivery mode (74, 76–78). A 2025 review noted that deep learning is commonly used for feature selection and diagnosis; several studies used neural networks for CTG analysis with strong performance (AUCs 0.87 to 0.987) (79–81). Random forest models were also common and useful for distinguishing normal from non-reassuring tracings (76, 78, 82, 83). One categorized delivery mode from CTG signals with an AUC of 0.967 (78). Of note, other ML models in this same study had varying performance; a decision tree model had an AUC of 0.839 while ensemble technique models had AUCs ranging from 0.949 to 0.955 (78). Another study modeled how neuraxial anesthesia and vasopressor dosing affected CTG signals, with a random forest estimating fetal heart rate changes with a mean standard error of 0.92 (82). AI tools have also begun clinical piloting, including remote fetal surveillance software that assessed neonatal outcomes and provider-perceived safety (84). With clinical deployment and larger training datasets, these models could improve intrapartum surveillance (84).

4.4 Imaging and radiomics

Fifteen articles (9%) addressed imaging and radiomics, with 8 solely in this domain and 7 spanning multiple domains including imaging and radiomics. The articles clustered into two themes. The first theme was MRI- and ultrasound-derived models for placenta accreta spectrum (PAS), discussed above (30), with complementary work applying deep learning radiomic analysis of MRI combined with clinical characteristics for PAS diagnosis and subtyping (AUC 0.85, n = 361) (10) and MRI radiomic prediction of hysterectomy in PAS (accuracy 0.92, n = 241) (11). The second theme was procedural ultrasound for obstetric anesthesia. Articles examined a software system for lumbar level identification (n = 76) (12), a case of AI-enhanced ultrasound for neuraxial placement in a severely obese parturient (85), and a neural network for real-time cricothyrotomy landmark identification using wireless handheld ultrasound (AUC 0.98, n = 608), which is relevant to failed-airway preparation on labor and delivery. Of note, different versions of the neural network model used had a range of performance with AUC values ranging from 0.96 to 0.98 (86).

The combination of radiomic features with clinical and laboratory data (fusion models) outperform either alone, suggesting the most useful applications will not rely on imaging data alone (10, 87). Apart from the small procedural ultrasound studies, none of the included imaging models have been evaluated prospectively in a workflow where the models’ predictions influenced operative planning or anesthetic selection.

4.5 Patient education and conversational AI

Fifteen articles (9%) addressed patient education and conversational AI, the most rapidly emerging domain, with 15 of 16 studies published in 2024 or 2025 (Figure 3), tracking the public release of large language models (88). Most studies assessed the accuracy, readability, and quality of large language model (LLM) responses to common questions about labor epidural analgesia. Multiple 2025 studies evaluated ChatGPT and other chatbots with consistent findings: responses were generally accurate and appropriately framed but often exceeded the recommended sixth-to-eighth grade reading level (89), and responses occasionally included factual errors or omitted clinically important caveats (90–92). This matters because patients may rely on chatbot responses, and incomplete or incorrect information can cause maternal or fetal harm. A comparative analysis of LLMs for epidural analgesia education found meaningful variation in accuracy and appropriateness (93), and a structured assessment evaluated LLMs as a tool for overcoming language barriers in obstetric anesthesia (94). A 2025 evaluation of ChatGPT-5 in obstetric and gynecologic emergencies (15 scenarios) highlighted reliability gaps that persist in the newest models (90).

Two applications point to more sophisticated uses. First, the “Mothers’ Assistant for Labor analgesia (MALA)” is an interactive avatar-based tool for obstetric anesthesia patient education (95). A 2023 survey of pregnant patients (n = 349) explored their perspectives on AI in clinical care, complementing clinician-centered evaluations (96).

This domain holds clear potential as a supplement to, not a replacement for, informed consent discussions. However, no included study prospectively evaluated whether LLM-supported education changed patient knowledge, anxiety, or satisfaction versus standard counseling. Beyond assessment of efficacy, real-world deployment of these models must also urgently prioritize improved readability, considerations of health literacy, and multi-lingual validation.

4.6 Provider education and simulation

Only two studies addressed provider education and simulation as a primary AI domain, a surprising gap given simulation's prominence in anesthesiology training (97). A 2025 study compared four LLMs (ChatGPT-4o, Claude 3.5 Sonnet, DeepSeek-R1, and Grok 3) on complex anesthesia decision-making across 30 high-risk scenarios (10 obstetric, 20 geriatric) (98). All four generated structured anesthesia plans, though performance varied; DeepSeek-R1 performed best in the obstetric subgroup, followed by ChatGPT-4o and Claude 3.5 Sonnet, with strengths in decision logic and contingency planning. This suggests that LLMs may aid case-based reasoning and simulation preparation, though the study was limited by small sample size, single-round prompting, and no real-time or outcome-based validation.

A 2025 review described simulation, digital education, telemedicine, and AI for peripartum team training in neonatal resuscitation (99). Noting that resuscitation skills can decline within months, the authors frame AI as an emerging but investigational tool for predicting deterioration, guiding management, and enabling virtual training through digital twin scenarios.

Together, these studies suggest AI's most immediate role is structured case generation, decision-support rehearsal, protocol reinforcement, and simulation of rare, high-risk emergencies. AI-augmented simulation for scenarios such as maternal cardiac arrest, massive postpartum hemorrhage, failed neuraxial conversion, and difficult airway management remain promising but underdeveloped.

4.7 AI in anesthesia training and workforce

Three studies examined the obstetric anesthesiologist as the subject, rather than the user, of AI. Two addressed generative AI in fellowship personal statements, both their creation by applicants and evaluation by program directors, finding that evaluators could not reliably distinguish LLM-generated from human-authored text, raising concerns about deception and the validity of the application process (100, 101). The third applied neural networks to identify burnout among obstetric anesthesiologists, finding contributors including workload, lack of work-life balance, emotional exhaustion, and job dissatisfaction (102). Though few in number, these studies may signal growing interest in physician-centered research.

4.8 Existing challenges and future directions

Four priorities recur across the eight domains surveyed. First, external validation should become the norm: only 14% of the included 154 original research studies reported external validation in an independent dataset. The documented drop in postpartum hemorrhage prediction when models move between health systems, including a C-statistic falling from 0.93 to 0.57 in one study (27), underscores the risk of relying on internal validation (103) Multi-site development, prespecified external validation cohorts, and adherence to standards such as TRIPOD-AI would substantially advance the field (104).

Second, predictive performance is necessary but not sufficient for deployment. Across every domain, a gap persisted between reported AUC and demonstrated effects on clinical decisions, workflow, or maternal-neonatal outcomes. Most of the 154 original research studies were retrospective development or internal validation; prospective evaluations of changes in clinician behavior or patient outcomes were rare. The nine registered clinical trials identified are an important early step, but have narrow topics: labor analgesia and its sequelae, lung ultrasound interpretation, automated spinal landmark identification, fetal heart rate analysis during neuraxial analgesia, and individualized risk scoring for placenta previa and delivery outcomes (105–112). The few studies show how limited the trial pipeline remains relative to model development activity.

Third, integration into obstetric anesthesia workflow was largely unaddressed. Few studies described how a model's output would reach the anesthesiologist, when, through which interface, and with what decision support. A risk score delivered at admission differs from one that updates during labor or interrupts with an alert. These deployment concerns: interface design, alert burden, electronic health record integration, accountability for AI-influenced decisions, and equity of access across high- and low-resource settings will determine whether technical progress translates into better maternal care. Successful clinical adoption will depend upon a research and implementation framework that includes external model validation, prospective trials, thorough integration into electronic health records, clinician feedback, and favorable outcome assessments.

Finally, the utilization of AI in maternal healthcare has significant ethical and regulatory implications. First, algorithmic biases can exacerbate existing health inequities (113–116). Algorithmic biases can arise from unbalanced training datasets, historical health inequities, developer biases, race-based clinical adjustments, and “black-box” model methodology (116). AI clinical tools must be developed with diverse datasets, rigorous training, transparent auditing, and interdisciplinary input to ensure equity (113, 115, 116). Adopting frameworks throughout the lifecycle of AI model development and implementation is another strategy to address bias (116). Additionally, protecting patient privacy and data security and ensuring patient consent are of paramount importance (113). Multi-disciplinary oversight committees and clear ethical and regulatory guidelines are needed to safeguard patient rights and information and guarantee accountability (113).

4.9 Study strengths and limitations

This review has several strengths. The search combined controlled vocabulary with comprehensive free-text terms across two major databases, identified 3,135 records, and included 172 studies (of which 154 were original research studies, 17 were systematic or narrative reviews, 1 case report)), making it, to our knowledge, the largest scoping review of AI in obstetric anesthesia to date. Selection used two reviewers with consensus adjudication, and charting was structured by AI domain, ML class, intended application, validation approach, and aspect of care, enabling systematic comparison across heterogeneous studies.

Several limitations apply. First, the search was restricted to English-language publications, potentially underrepresenting regions where practice and AI priorities differ. Second, the search strategy was limited to two databases (PubMed and Embase) due to their broad indexing of articles and clinical relevance. Other databases such as Scopus, Web of Science, or IEEE Xplore were not used, and we acknowledge that there are articles that may have been missed from these databases due to our search strategy. Next, as a bridge search was not run between PubMed and Embase searches (March vs April 2026), some articles may have been missed. Third, scoping review methodology includes no formal risk-of-bias assessment, so we do not report study quality beyond descriptive characteristics (117). Fourth, the boundary between obstetric anesthesia and adjacent specialties is not clean, so some domain assignments required judgment. Fifth, studies were compared using the best-performing models reported from each study (using AUC, AUROC, accuracy) which may lead to an overly optimistic assessment of model performance. This was done to more clearly facilitate comparisons among a broad range of studies. Where appropriate, other poorer performing models from each study were included for a more balanced assessment of model performance. Finally, excluding preprints, conference abstracts, and editorials may underrepresent the most recent work in a fast-moving field.

5 Conclusion

AI applications in obstetric anesthesia have expanded rapidly over the past decade, with most work appearing in the last three years across seven domains, from postpartum hemorrhage and placenta accreta spectrum prediction to large language model-driven patient education and workforce-facing tools. Reported discrimination is frequently strong, with AUCs commonly 0.75 to 0.95, yet only a minority of studies have been externally validated, fewer still prospectively evaluated, and almost none embedded into obstetric anesthesia workflow. The limited work on provider education and simulation suggests AI may also aid training through case-based reasoning, rehearsal of rare high-risk scenarios, protocol reinforcement, and neonatal resuscitation education, though these uses remain preliminary and require prospective validation. Realizing AI's potential to reduce preventable maternal morbidity and mortality will require a deliberate shift from model development toward external validation, prospective evaluation, and responsible adoption and integration, with obstetric anesthesiologists serving as designers, evaluators, and stewards of the tools that will increasingly shape their practice.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fanes.2026.1896595/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: André Van Zundert, The University of Queensland, Australia

Reviewed by: Elhadi Miskeen, University of Bisha, Saudi Arabia

Ugo Maria Pierucci, Ospedale dei Bambini Vittore Buzzi, Italy

Charles Onyeka Nwamekwe, Nnamdi Azikiwe University, Nigeria

Statements

Conflict of interest

PAG is a co-inventor on patents related to the development of novel alkylphenols for the treatment of neuropathic pain and serves on the Scientific Advisory Board for Akelos (New York, NY), a research-based biotechnology company that has secured a licensing agreement for the use of those patents. The work presented here is fully independent of those efforts.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author RW declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.

Funding

The author(s) declared that financial support was not received for this work and/or its publication.

Author contributions

MG: Writing – original draft, Investigation, Data curation, Validation, Formal analysis, Methodology, Writing – review & editing. ER: Validation, Investigation, Data curation, Writing – review & editing, Methodology, Writing – original draft, Formal analysis. RJ: Methodology, Validation, Conceptualization, Supervision, Writing – review & editing. PG: Project administration, Methodology, Validation, Writing – review & editing, Supervision. JR: Supervision, Methodology, Project administration, Writing – review & editing, Validation. RW: Conceptualization, Validation, Writing – review & editing, Investigation, Resources, Supervision, Methodology, Writing – original draft, Project administration.

References

  1. 1. Loftus TJ, Tighe PJ, Filiberto AC, Efron PA, Brakenridge SC, Mohr AM. Artificial intelligence and surgical decision-making. JAMA Surg. (2020) 155 (2): 148 – 58. doi:10.1001/jamasurg.2019.4917
  2. 2. Hashimoto DA, Witkowski E, Gao L, Meireles O, Rosman G. Artificial intelligence in anesthesiology: current techniques, clinical applications, and limitations. Anesthesiology. (2020) 132 (2): 379 – 94. doi:10.1097/aln.0000000000002960
  3. 3. Rajkomar A, Dean J, Kohane I. Machine learning in medicine. N Engl J Med. (2019) 380 (14): 1347 – 58. doi:10.1056/nejmra1814259
  4. 4. Ghahramani Z. Probabilistic machine learning and artificial intelligence. Nature. (2015) 521 (7553): 452 – 9. doi:10.1038/nature14541
  5. 5. Rozental O, White RS. Anesthesia information management systems: evolution of the paper anesthetic record to a multisystem electronic medical record network that streamlines perioperative care. J Anesth Hist. (2019) 5 (3): 93 – 8. doi:10.1016/j.janh.2019.04.001
  6. 6. Elmaleh Y, Guessous K, Delvaux BV. Precision perioperative AI: from signals, images, and records to applications in anesthesia-A narrative mini-review proposing an operational framework. Front Med (Lausanne). (2026) 13:1811197. doi:10.3389/fmed.2026.1811197
  7. 7. Kovacheva VP, Burns ML. Artificial intelligence and predictive analytics in obstetric anesthesia: early warning for maternal complications. Curr Opin Anaesthesiol. (2026) 39 (3): 231 – 9. doi:10.1097/aco.0000000000001613
  8. 8. Haug CJ, Drazen JM. Artificial intelligence and machine learning in clinical medicine, 2023. N Engl J Med. (2023) 388 (13): 1201 – 8. doi:10.1056/nejmra2302038
  9. 9. Lipschuetz M, Guedalia J, Cohen SM, Unger R, Yagel S, Sompolinsky Y. Studies in health technology and informatics. Stud Health Technol Inform. (2024) 315: 3 – 7. doi:10.3233/shti240096
  10. 10. Zheng C, Zhong J, Wang Y, Cao K, Zhang C, Yue P. Deep learning radiomic analysis of MRI combined with clinical characteristics diagnoses placenta accreta spectrum and its subtypes. J Magn Reson Imaging. (2024) 60 (6): 2705 – 15. doi:10.1002/jmri.29317
  11. 11. Leitch K, Shahedi M, Dormer JD, Do QN, Xi Y, Lewis MA. Placenta accreta spectrum and hysterectomy prediction using MRI radiomic features. Medical Imaging 2022: Computer-Aided Diagnosis. Proc SPIE Int Soc Opt Eng. (2022): 120331. doi:10.1117/12.2611587
  12. 12. Hetherington J, Brohan J, Rohling R, Gunka V, Abolmaesumi P, Albert A. A novel ultrasound software system for lumbar level identification in obstetric patients. Can J Anesth. (2022) 69 (10): 1211 – 9. doi:10.1007/s12630-022-02300-6
  13. 13. Chen Y, Shiels MS, Uribe-Leitz T, Molina RL, Lawrence WR, Freedman ND. Pregnancy-related deaths in the US, 2018−2022. JAMA Netw Open. (2025) 8 (4): e254325. doi:10.1001/jamanetworkopen.2025.4325
  14. 14. Khusid E, Lui B, Ibarra A, Villegas K, White RS. Review of racial/ethnic disparities in obstetrics-related anesthesia administration and pain management. Pain Manag. (2023) 13 (7): 415 – 22. doi:10.2217/pmt-2023-0034
  15. 15. White R, Lui B, Bryant-Huppert J, Chaturvedi R, Hoyler M, Aaronson J. Economic burden of maternal mortality in the USA, 2018−2020. J Comp Eff Res. (2022) 11 (13): 927 – 33. doi:10.2217/cer-2022-0056
  16. 16. White R, Aaronson J. Obstetric and perinatal racial and ethnic disparities. Curr Opin Anaesthesiol. (2022) 35 (3): 260 – 6. doi:10.1097/ACO.0000000000001133
  17. 17. Bienstock JL, Eke AC, Hueppchen NA. Postpartum hemorrhage. N Engl J Med. (2021) 384 (17): 1635 – 45. doi:10.1056/NEJMra1513247
  18. 18. Kumaraswami S, Butwick A. Latest advances in postpartum hemorrhage management. Best Pract Res Clin Anaesthesiol. (2022) 36 (1): 123 – 34. doi:10.1016/j.bpa.2022.02.004
  19. 19. Lengerich B, Caruana R, Painter I, Weeks W, Sitcov K, Souter V. Interpretable machine learning predicts postpartum hemorrhage with severe maternal morbidity in a lower-risk laboring obstetric population. Am J Obstet Gynecol MFM. (2024) 6 (8): 101391. doi:10.1016/j.ajogmf.2024.101391
  20. 20. Mehrnoush V, Ranjbar A, Farashah MV, Darsareh F, Shekari M, Jahromi MS. Prediction of postpartum hemorrhage using traditional statistical analysis and a machine learning approach. AJOG Glob Rep. (2023) 3 (2): 100185. doi:10.1016/j.xagr.2023.100185
  21. 21. Ranjbar A, Ghamsari SR, Boujarzadeh B, Mehrnoush V, Darsareh F. Predicting risk of postpartum hemorrhage using machine learning approach: a systematic review. Gynecol Obstet Clin Med. (2023) 3 (3): 170 – 4. doi:10.1016/j.gocm.2023.07.002
  22. 22. Woo EG, Zighelboim I, Gifford T, Bell JG, Milthorpe H, Alsentzer E. Predicting postpartum hemorrhage using clinical features extracted with large language models. O G Open. (2025) 2 (5): e128. doi:10.1097/og9.0000000000000128
  23. 23. Alsentzer E, Rasmussen MJ, Fontoura R, Cull AL, Beaulieu-Jones B, Gray KJ. Zero-shot interpretable phenotyping of postpartum hemorrhage using large language models. NPJ Digit Med. (2023) 6 (1): 212. doi:10.1038/s41746-023-00957-x
  24. 24. Wakefield B, Zapf M, Ende H. Artificial intelligence in prediction of postpartum hemorrhage: a primer and review. Int J Obstet Anesth. (2025) 63: 104694. doi:10.1016/j.ijoa.2025.104694
  25. 25. Dogru S, Ezveci H, Akkus F, Bahçeci P, Karanfil YF, Acar A. Artificial intelligence in predicting postpartum hemorrhage in twin pregnancies undergoing cesarean section. Twin Res Hum Genet. (2025) 28 (1): 53 – 9. doi:10.1017/thg.2024.48
  26. 26. Gong J, Chen Z, Zhang Y, Liu Y-, Pu J-, Xiong C- Risk-factor model for postpartum hemorrhage after cesarean delivery: a retrospective study based on 3498 patients. Sci Rep. (2022) 12 (1):22100. doi:10.1038/s41598-022-23636-5
  27. 27. Meyer SR, Carver A, Joo H, Venkatesh KK, Jelovsek JE, Klumpner TT. External validation of postpartum hemorrhage prediction models using electronic health record data. Am J Perinatol. (2024) 41 (05): 598 – 605. doi:10.1055/a-1745-1348
  28. 28. Bonanni G, Lopez-Giron MC, Allen L, Fox K, Silver RM, Hobson SR. Guidelines on placenta accreta spectrum disorders. JAMA Netw Open. (2025) 8 (7): e2521909. doi:10.1001/jamanetworkopen.2025.21909
  29. 29. Jauniaux E, Bunce C, Grønbeck L, Langhoff-Roos J. Prevalence and main outcomes of placenta accreta spectrum: a systematic review and meta-analysis. Am J Obstet Gynecol. (2019) 221 (3): 208 – 18. doi:10.1016/j.ajog.2019.01.233
  30. 30. Hu Y, Liu T, Pang S, Ling X, Wang Z, Li W. Deep learning-assisted diagnosis of placenta accreta spectrum using the DenseNet-121 model: a multicenter, retrospective study. J Imaging Inform Med. (2025) 38 (6): 3899 – 908. doi:10.1007/s10278-025-01475-w
  31. 31. Zhang Y, Hu M, Wen X, Huang Y, Luo R, Chen J. MRI-based radiomics nomogram in patients with high-risk placenta accreta spectrum: can it aid in the prenatal diagnosis of intraoperative blood loss? Abdom Radiol (NY). (2023) 48 (3): 1107 – 18. doi:10.1007/s00261-022-03784-y
  32. 32. Li M, Su X, Liao W, Huang L, Yang Y, Wu X. Development and validation of an interpretable machine learning-based prediction model of postpartum hemorrhage in placenta previa following cesarean section: a multicenter study. Reprod Sci. (2025) 32 (9): 3062 – 73. doi:10.1007/s43032-025-01937-0
  33. 33. Li Y, Li L, Song X, Meng F, Zhang M, Li Y. Development of a predictive model for severe peripartum hemorrhage in placenta accreta spectrum cases under neuraxial anesthesia: a multicenter retrospective analysis. Ther Adv Reprod Health. (2025) 19:26334941251317644. doi:10.1177/26334941251317644
  34. 34. Xu J, Shao Q, Chen R, Xuan R, Mei H, Wang Y. A dual-path neural network fusing dual-sequence magnetic resonance image features for detection of placenta accrete spectrum (PAS) disorder. Math Biosci Eng. (2022) 19 (6): 5564 – 75. doi:10.3934/mbe.2022260
  35. 35. Shazly SA, Hortu I, Shih J-C, Melekoglu R, Fan S, Ahmed FA. Prediction of clinical outcomes in women with placenta accreta spectrum using machine learning models: an international multicenter study. J Matern Fetal Neonatal Med. (2022) 35 (25): 6644 – 53. doi:10.1080/14767058.2021.1918670
  36. 36. Henderson JT, Webber EM, Thomas RG, Vesco KK. Screening for hypertensive disorders of pregnancy: US preventive services task force final recommendation statement. JAMA. (2023) 330 (11): 1074 – 82.
  37. 37. Malik V, Agrawal N, Prasad S, Talwar S, Khatuja R, Jain S. Prediction of preeclampsia using machine learning: a systematic review. Cureus. (2024) 16 (12):e76095. doi:10.7759/cureus.76095
  38. 38. Edvinsson C, Björnsson O, Erlandsson L, Hansson SR. Predicting intensive care need in women with preeclampsia using machine learning – A pilot study. Hypertens Pregnancy. (2024) 43 (1):2312165. doi:10.1080/10641955.2024.2312165
  39. 39. Arany Z. Peripartum cardiomyopathy. N Engl J Med. (2024) 390 (2): 154 – 64. doi:10.1056/NEJMra2306667
  40. 40. Banayan J, Rana S, Mueller A, Tung A, Ramadan H, Arany Z. Cardiogenic shock in pregnancy: analysis from the national inpatient sample. Hypertens Pregnancy. (2017) 36 (2): 117 – 23. doi:10.1080/10641955.2016.1242606
  41. 41. Adedinsewo DA, Morales-Lara AC, Afolabi BB, Kushimo OA, Mbakwem AC, Ibiyemi KF. Artificial intelligence guided screening for cardiomyopathies in an obstetric population: a pragmatic randomized clinical trial. Nat Med. (2024) 30 (10): 2897 – 906. doi:10.1038/s41591-024-03243-9
  42. 42. Chaemsaithong P, Sahota DS, Poon LC. First trimester preeclampsia screening and prediction. Am J Obstet Gynecol. (2022) 226 (2S): S1071 – 97. doi:10.1016/j.ajog.2020.07.020
  43. 43. Salman Zahid M, Shikha Jha M, Gurleen Kaur M, Youn-Hoa Jung M, Anum S, Minhas M. PARCCS: a machine learning risk-prediction model for acute peripartum cardiovascular complications during delivery admissions. JACC Adv. (2024) 3 (8): 101095. doi:10.1016/j.jacadv.2024.101095
  44. 44. Patel S, Ciechanowicz S, Blumenfeld Y, Sultan P. Epidural-related maternal fever: incidence, pathophysiology, outcomes, and management. Am J Obstet Gynecol. (2023) 228 (5S): S1283 – 304. doi:10.1016/j.ajog.2022.06.026
  45. 45. Wang H, Yang Z, Wei S, Xia L, Li Y, Wu X. Perinatal outcomes and risk factors for epidural analgesia-associated intrapartum maternal fever: a retrospective study. J Matern Fetal Neonatal Med. (2023) 36 (1): 2179383. doi:10.1080/14767058.2023.2179383
  46. 46. Jiang H, Li N. Machine learning-driven intrapartum fever prediction: a comprehensive large retrospective cohort study integrating inflammatory and obstetric markers. Int J Gen Med. (2025) 18: 6949 – 60. doi:10.2147/IJGM.S550437
  47. 47. Zhang G, Yang Y, An R, Tan Z. Development of a machine learning model to predict epidural-related maternal fever during labor analgesia: a multi-algorithm comparative study with prospective implementation framework. Int J Women’s Health. (2025) 17: 5439 – 51. doi:10.2147/IJWH.S560693
  48. 48. Liu B, Ling L, Jia F, Wei D, Li H, Li Y. Development and validation of a machine learning model for predicting intrapartum fever using pre-labor analgesia clinical indicators: a multicenter retrospective study. BMC Pregnancy Childbirth. (2025) 25 (1):243. doi:10.1186/s12884-025-07203-0
  49. 49. Guo X, Zhang H, Mei H. Machine learning algorithms to predict epidural-related maternal fever: a retrospective study. Front Pharmacol. (2025) 16:1614770. doi:10.3389/fphar.2025.1614770
  50. 50. Gratz I, Baruch M, Takla M, Seaman J, Allen I, McEniry B. The application of a neural network to predict hypotension and vasopressor requirements non-invasively in obstetric patients having spinal anesthesia for elective cesarean section (C/S). BMC Anesthesiol. (2020) 20 (1):98. doi:10.1186/s12871-020-01015-9
  51. 51. Tan HS, Liu N, Sultana R, Han N-LR, Tan CW, Zhang J. Prediction of breakthrough pain during labour neuraxial analgesia: comparison of machine learning and multivariable regression approaches. Int J Obstet Anesth. (2021) 45: 99 – 110. doi:10.1016/j.ijoa.2020.08.010
  52. 52. Abdelgadir Elhabeeb SM, Mahmoud Ali SH, Ahmed Elkhidir Babikir MM, Abdalla Mohammed FS, Mahmoud Ali SH, Abd Elfrag Mohamed NA. Enhancing obstetric decision-making with AI: a systematic review of AI models for predicting mode of delivery. Cureus. (2025) 17 (5):e83655. doi:10.7759/cureus.83655
  53. 53. Yang M, Long D, Li Y, Liu X, Bai Z, Li Z. An explainable machine learning model in predicting vaginal birth after cesarean section. J Matern Fetal Neonatal Med. (2025) 38 (1):2546544. doi:10.1080/14767058.2025.2546544
  54. 54. Zoha S, Alam S, Sifat I, Sultana N, Kibria M. Identifying determinants and predicting cesarean section delivery among Bangladeshi women using machine learning: insight from BDHS 2022 data. PLOS Global Public Health. (2025) 5 (11): e0005494. doi:10.1371/journal.pgph.0005494
  55. 55. Givon I, Bor N, Matot R, Friedrich L, Gross D, Konforty G. Dynamic machine learning models for predicting cesarean delivery risk in women with no prior cesarean delivery: a retrospective nationwide cohort analysis. Int J Gynecol Obstet. (2025) 171 (2): 775 – 83. doi:10.1002/ijgo.70234
  56. 56. Schmauder S, Sandström A, Boman M, Martin C, Stephansson O. Machine learning prediction models for mode of delivery in prolonged pregnancies in Sweden. Sci Rep. (2025) 15 (1):32487. doi:10.1038/s41598-025-19198-x
  57. 57. Bai J, Kang X, Wang W, Yang Z, Ou W, Huang Y. A multimodal model in the prediction of the delivery mode using data from a digital twin-empowered labor monitoring system. Digit Health. (2024) 10:20552076241304934. doi:10.1177/20552076241304934
  58. 58. Wong MS, Wells M, Zamanzadeh D, Akre S, Pevnick JM, Bui AAT. Applying automated machine learning to predict mode of delivery using ongoing intrapartum data in laboring patients. Am J Perinatol. (2024) 41: e412 – 9. doi:10.1055/a-1885-1697
  59. 59. Lodi M, Poterie A, Exarchakis G, Brien C, Lafaye de Micheaux P, Deruelle P. Prediction of cesarean delivery in class III obese nulliparous women: an externally validated model using machine learning. J Gynecol Obstet Hum Reprod. (2023) 52 (7): 102624. doi:10.1016/j.jogoh.2023.102624
  60. 60. Awawdeh S, Rawashdeh H, Aljalodi H, Shamleh R, Alshorman S. Vaginal birth after cesarean section prediction model for Jordanian population. Comput Biol Chem. (2023) 104: 107877. doi:10.1016/j.compbiolchem.2023.107877
  61. 61. Huang X, Di X, Lin S, Yao M, Zheng S, Liu S. Artificial intelligence-based prediction of second stage duration in labor: a multicenter retrospective cohort analysis. eClinicalMed. (2025) 80 (103072): 103072. doi:10.1016/j.eclinm.2025.103072
  62. 62. American College of Obstetricians and Gynecologists. First and second stage labor management: aCOG clinical practice guideline no. 8. Obstet Gynecol. (2024) 143 (1): 144 – 62. doi:10.1097/AOG.0000000000005447
  63. 63. Hamilton ARL, Drzymalski D. A retrospective review of clinical and non-clinical factors associated with unscheduled cesarean deliveries at one urban medical center. Int J Gynecol Obstet. (2022) 157 (1): 154 – 8. doi:10.1002/ijgo.13727
  64. 64. Yang XJ, Sun SS. Comparison of maternal and fetal complications in elective and emergency cesarean section: a systematic review and meta-analysis. Arch Gynecol Obstet. (2017) 296 (3): 503 – 12. doi:10.1007/s00404-017-4445-2
  65. 65. Butwick A, Baer RJ, Farooqi N, Stephansson O, Jelliffe-Pawlowski LL. Severe perioperative surgical morbidity with cesarean delivery. Obstet Gynecol. (2026) 147 (6): 838 – 47. doi:10.1097/AOG.0000000000006041
  66. 66. Meyer R, Hendin N, Zamir M, Mor N, Levin G, Sivan E. Implementation of machine learning models for the prediction of vaginal birth after cesarean delivery. J Matern Fetal Neonatal Med. (2022) 35 (19):3677–83. doi:10.1080/14767058.2020.1837769
  67. 67. Hu Y, Zhang X, Slavin V, Enticott J, Callander E. Explainable machine learning model for predicting cesarean section following induction of labor: development and external validation using real-world data. PLOS Digit Health. (2025) 4 (11): e0001061. doi:10.1371/journal.pdig.0001061
  68. 68. American College of Obstetricians and Gynecologists. ACOG practice bulletin no. 205: vaginal birth after cesarean delivery. Obstet Gynecol. (2019) 133 (2): e110 – 27. doi:10.1097/AOG.0000000000003078
  69. 69. Grobman WA, Sandoval G, Rice MM, Bailit JL, Chauhan SP, Costantine MM. Prediction of vaginal birth after cesarean delivery in term gestations: a calculator without race and ethnicity. Am J Obstet Gynecol. (2021) 225 (6): 664.e1 – 7. doi:10.1016/j.ajog.2021.05.021
  70. 70. Malvasi A, Malgieri LE, Stark M, Tinelli A. Dystocia, delivery, and artificial intelligence in labor management: perspectives and future directions. J Clin Med. (2024) 13 (21): 6410. doi:10.3390/jcm13216410
  71. 71. Tsur A, Batsry L, Toussia-Cohen S, Rosenstein M, Barak O, Brezinov Y. Development and validation of a machine-learning model for prediction of shoulder dystocia. Ultrasound Obstet Gynecol. (2020) 56 (4): 588 – 96. doi:10.1002/uog.21878
  72. 72. Malvasi A, Malgieri LE, Stark M, Di Naro E, Farine D, Baldini GM. The contribution of AIDA (artificial intelligence dystocia algorithm) to cesarean section within robson classification group. J Imaging. (2025) 11 (8): 276. doi:10.3390/jimaging11080276
  73. 73. American College of Obstetricians and Gynecologists. ACOG clinical practice guideline no. 10: intrapartum fetal heart rate monitoring: interpretation and management. Obstet Gynecol. (2025) 146 (4): e1 – 27. doi:10.1097/AOG.0000000000006049
  74. 74. Lovers A, Daumer M, Frasch MG, Ugwumadu A, Warrick P, Vullings R. Advancements in fetal heart rate monitoring: a report on opportunities and strategic initiatives for better intrapartum care. BJOG. (2025) 132 (7): 853 – 66. doi:10.1111/1471-0528.18097
  75. 75. Pinas A, Chandraharan E. Continuous cardiotocography during labour: analysis, classification and management. Best Prac Res Clin Obstet Gynaecol. (2016) 30: 33 – 47. doi:10.1016/j.bpobgyn.2015.03.022
  76. 76. Vargas-Calixto J, W. Wu Y, Kuzniewicz M, Cornet M-C, Forquer H, Gerstley L. Prediction of hypoxic-ischemic encephalopathy using events in fetal heart rate and uterine pressure. Comput Cardiol (2010). (2023) 50: 1 – 4. doi:10.22489/cinc.2023.380
  77. 77. Esteban-Escaño J, Castán B, Castán S, Chóliz-Ezquerro M, Asensio C, Laliena AR. Machine learning algorithm to predict acidemia using electronic fetal monitoring recording parameters. Entropy. (2021) 24 (1): 68. doi:10.3390/e24010068
  78. 78. Ricciardi C, Improta G, Amato F, Cesarelli G, Romano M. Classifying the type of delivery from cardiotocographic signals: a machine learning approach. Comput Methods Programs Biomed. (2020) 196 (105712): 105712. doi:10.1016/j.cmpb.2020.105712
  79. 79. Zhao Z, Zhang Y, Comert Z, Deng Y. Computer-aided diagnosis system of fetal hypoxia incorporating recurrence plot with convolutional neural network. Front Physiol. (2019) 10:255. doi:10.3389/fphys.2019.00255
  80. 80. Fergus P, Hussain A, Al-Jumeily D, Huang DS, Bouguila N. Classification of caesarean section and normal vaginal deliveries using foetal heart rate signals and advanced machine learning algorithms. Biomed Eng Online. (2017) 16 (1): 89. doi:10.1186/s12938-017-0378-z
  81. 81. Iqbal Y, Hussain S, Ayoub M, Idrees MF, Munir S, Rafiq U. E-NN: fetal heart rate analysis during pregnancy using an ensemble densenet-BC and convolutional neural network-based framework. J Popul Ther Clin Pharmacol. (2024) 31 (9): 669 – 81. doi:10.53555/az0c9m30
  82. 82. Riveros-Perez E, Polania-Gutierrez JJ, Avella-Molano B. Fetal heart rate changes and labor neuraxial analgesia: a machine learning approach. BMC Pregnancy Childbirth. (2023) 23 (1): 329. doi:10.1186/s12884-023-05632-3
  83. 83. Silva LM, Joiner D, Qezelbash-Chamak J, Kim M, Garvan C, Berg M. Longitudinal fetal heart rate analysis identifies neonates with metabolic acidemia requiring therapeutic hypothermia. Am J Obstet Gynecol. (2026) 234 (6): 1690 – 9. doi:10.1016/j.ajog.2025.09.019
  84. 84. Lowery D, De Leon B, Krening C, Dempsey A, Dwork P, Brou L. OB HUB: remote electronic fetal monitoring surveillance. MCN Am J Matern Child Nurs. (2023) 48 (2): 62 – 8. doi:10.1097/NMC.0000000000000891
  85. 85. Compagnone C, Borrini G, Calabrese A, Taddei M, Bellini V, Bignami E. Artificial intelligence-enhanced ultrasound in a severely obese parturient: a case report. Ultrasound J. (2022) 14 (1): 34. doi:10.1186/s13089-022-00283-5
  86. 86. Wu C-Y, Li J-D, Shih P-Y, Huang C-C, Cheng H-L, Wu C-Y. Real-time identification of cricothyrotomy landmarks in emergency care and obstetric patients using wireless handheld ultrasound and edge-computing artificial intelligence: a prospective observational study. J Med Syst. (2025) 49 (1):131. doi:10.1007/s10916-025-02275-z
  87. 87. Zheng C, Yue P, Cao K, Wang Y, Zhang C, Zhong J. Predicting intraoperative blood loss during cesarean sections based on multi-modal information: a two-center study. Abdominal Radiol. (2024) 49 (7): 2325 – 39. doi:10.1007/s00261-024-04419-0
  88. 88. Le NL, Abel MH. How well do LLMs predict prerequisite skills? Zero-shot comparison to expert-defined concepts. arXiv. (2025):7245–50. doi:10.1109/SMC58881.2025.11343557
  89. 89. Alamleh S, Mavedatnia D, Francis G, Le T, Davies J, Lin V. Readability, reliability, and quality analysis of internet-based patient education materials and large language models on meniere disease. J Otolaryngol Head Neck Surg. (2025) 54:19160216251360651. doi:10.1177/19160216251360651
  90. 90. Arslan H, Arslan K, Gun M, Karkin P, Arslanoglu T. Evaluation of ChatGPT-5 responses in obstetric and gynecological emergencies: concordance, readability, and clinical reliability. BMC Emerg Med. (2025) 25 (1):220. doi:10.1186/s12873-025-01396-5
  91. 91. Gondode P, Singh R, Mehta S, Singh S, Kumar S, Nayak S. Artificial intelligence chatbots versus traditional medical resources for patient education on labor epidurals: an evaluation of accuracy, emotional tone, and readability. Int J Obstet Anesth. (2025) 61: 104302. doi:10.1016/j.ijoa.2024.104302
  92. 92. Fiol AG, Mootz A, He Z, Delgado C, Ortiz V, Reale S. Accuracy of Spanish and English-generated ChatGPT responses to commonly asked patient questions about labor epidurals: a survey-based study among bilingual obstetric anesthesia experts. Int J Obstet Anesth. (2025) 61: 104290. doi:10.1016/j.ijoa.2024.104290
  93. 93. Patel A, Joseph S, Bhardwaj A, Bajaj P, Aggarwal I, Bailey C. A patient-focused comparative analysis of large language models for epidural analgesia education. A A Prac. (2025) 19 (8): e02025. doi:10.1213/XAA.0000000000002025
  94. 94. Lomas A, Broom M. Large language models for overcoming language barriers in obstetric anaesthesia: a structured assessment. Int J Obstet Anesth. (2024) 60: 104249. doi:10.1016/j.ijoa.2024.104249
  95. 95. Suri A, Sindwani G. Mothers’ assistant for labor analgesia (MALA): a novel artificial intelligence interactive avatar for patient education in obstetric anesthesia. Int J Obstet Anesth. (2025) 64: 104762. doi:10.1016/j.ijoa.2025.104762
  96. 96. Armero W, Gray K, Fields K, Cole N, Bates D, Kovacheva V. A survey of pregnant patients’ perspectives on the implementation of artificial intelligence in clinical care. J Am Med Inform Assoc. (2022) 30 (1): 46 – 53. doi:10.1093/jamia/ocac200
  97. 97. Yunoki K, Sakai T. The role of simulation training in anesthesiology resident education. J Anesth. (2018) 32 (3): 425 – 33. doi:10.1007/s00540-018-2483-y
  98. 98. Ruan Q, Shi J, Dai Y, Yang P, Zhu N, Wang S. Performance of large language models in complex anesthesia decision-making: a comparative study of four LLMs in high-risk patients. J Med Syst. (2025) 49 (1):122. doi:10.1007/s10916-025-02247-3
  99. 99. Fantin R, Wallner B, Lichtenberger P, Putzer G, Neubauer V, Griesmaier E. Advances in neonatal resuscitation for the obstetric anesthesiologist. Curr Opin Anesthesiol. (2025) 38 (3): 208 – 14. doi:10.1097/ACO.0000000000001462
  100. 100. Matsubara S. Rethinking personal statements submitted for obstetric anesthesia fellowship: beyond human-written or artificial intelligence-generated. Int J Obstet Anesth. (2025) 61: 104309. doi:10.1016/j.ijoa.2024.104309
  101. 101. Ruiz A, Kraus M, Arendt K, Schroeder D, Sharpe E. Artificial intelligence-created personal statements compared with applicant-written personal statements: a survey of obstetric anesthesia fellowship program directors in the United States. Int J Obstet Anesth. (2025) 61: 104293. doi:10.1016/j.ijoa.2024.104293
  102. 102. Cascella M, Simonini A, Coluccia S, Bignami EG, Fiore G, Petrucci E. Different artificial neural networks for predicting burnout risk in Italian anesthesiologists. J Anesth Analg Crit Care. (2025) 5 (1):40. doi:10.1186/s44158-025-00255-w
  103. 103. Cabitza F, Campagner A, Soares F, García de Guadiana-Romualdo L, Challa F, Sulejmani A. The importance of being external: methodological insights for the external validation of machine learning models in medicine. Comput Methods Programs Biomed. (2021) 208: 106288. doi:10.1016/j.cmpb.2021.106288
  104. 104. Collins GS, Moons KGM, Dhiman P, Riley RD, Beam AL, Van Calster B. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. (2024) 385: e078378. doi:10.1136/bmj-2023-078378
  105. 105. KK Women’s and Children’s Hospital, National University of Singapore. Automated Spinal Landmark Identification to Improve Patient Safety and Efficacy During Neuraxial Anaesthesia Needle Insertion. ClinicalTrials.gov identifier: NCT03535155. (2018).
  106. 106. Dal NA; Munzur University. The Effect of Neuro Linguistic Programming on Pain Levels After Cesarean Delivery: A Randomized Controlled Trial. identifier: NCT05033405. (2021).
  107. 107. Centro di Ricerca Clinica Salentino. Study of Delivery Outcomes After AIDA (Artificial Intelligence Dystocia Algorithm) Analysis. identifier: NCT06664112. (2024).
  108. 108. Shazly SAM; Assiut University. Clinical Outcomes of Placenta Previa - An Individualized Scoring System. identifier: NCT06589180. (2024).
  109. 109. Yavuzel S; Hacettepe University. Development of an Artificial Intelligence Algorithm to Predict Hypotension Risk After Induction in Cesarean Sections with Spinal Anesthesia. identifier: NCT06158542. (2023).
  110. 110. Augusta University. Fetal Heart Rate Changes and Labor Neuraxial Analgesia: A Machine Learning Approach. identifier: [NCT number needed]. (2022).
  111. 111. Zhang X; Huazhong University of Science and Technology. Clinical Study on the Effect of Labor Analgesia on the Postpartum Recovery Outcome of Chinese Parturients. identifier: NCT05666024. (2022).
  112. 112. Gunaydin B; Gazi University. Evaluation of Artificial Intelligence in Defining Lung Ultrasound Findings in Anesthesia Management of Preeclampsia. identifier: NCT05487014. (2022).
  113. 113. McAdams RM, Green TL. Equitable artificial intelligence in obstetrics, maternal–fetal medicine, and neonatology. Obstet Gynecol. (2024) 143 (5): 627 – 32. doi:10.1097/aog.0000000000005563
  114. 114. Obermeyer Z, Powers B, Vogeli C, Mullainathan S. Dissecting racial bias in an algorithm used to manage the health of populations. Science. (2019) 366 (6464): 447 – 53. doi:10.1126/science.aax2342
  115. 115. Ferryman K, Mackintosh M, Ghassemi M. Considering biased data as informative artifacts in AI-assisted health care. N Engl J Med. (2023) 389 (9): 833 – 8. doi:10.1056/nejmra2214964
  116. 116. Adegunle F, Chhatwal K, Arab S, Alabdaljabar MS, Raslan MA, Sayed O. Bias and oversight in clinical AI: a review of decision support tools and equity frameworks. J Gen Intern Med. (2026) 41 (7): 1957 – 68. doi:10.1007/s11606-026-10229-5
  117. 117. Drucker A, Fleming P, Chan A. Research techniques made simple: assessing risk of bias in systematic reviews. J Invest Dermatol. (2016) 136 (11): e109 – 14. doi:10.1016/j.jid.2016.08.021
  118. 118. Tan CW, Koh JZ, Jin H, Han N-LR, Cheng S-M, Ta AWA. Machine learning approach to predict postoperative pain after spinal morphine administration during caesarean delivery. Heliyon. (2024) 10 (23): e40602. doi:10.1016/j.heliyon.2024.e40602
  119. 119. Iraji MS. Prediction of fetal state from the cardiotocogram recordings using neural network models. Artif Intell Med. (2019) 96: 33 – 44. doi:10.1016/j.artmed.2019.03.005
  120. 120. Liu J, Wu T, Peng Y, Luo R. Grade prediction of bleeding volume in cesarean section of patients with pernicious placenta previa based on deep learning. Front Bioeng Biotechnol. (2020) 8:343. doi:10.3389/fbioe.2020.00343
  121. 121. Meyer R, Weisz B, Eilenberg R, Tsadok MA, Uziel M, Sivan E. Utilizing machine learning to predict unplanned cesarean delivery. Int J Gynecol Obstet. (2022) 161 (1): 255 – 63. doi:10.1002/ijgo.14433
  122. 122. Kuanar A, Akbar A, Sujata P, Kar D. Deep neural network (DNN) modelling for prediction of the mode of delivery. Eur J Obstet Gynecol Reprod Biol. (2024) 297: 241 – 8. doi:10.1016/j.ejogrb.2024.04.012

Citation

Givens, M. F., et al. (2026). Artificial intelligence and obstetric anesthesia: a scoping review. Frontiers in Anesthesiology, Volume 5. https://doi.org/
Received
Accepted
Published
Volume
5 - 2026

Copyright

© 2026 Givens, Rodriguez Caceres, Jotwani, Rubin, Goldstein and White.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: Robert S. White

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

ReviewWhen ECMO can no longer bridge: a narrative review and proposed consent templateAugust Felix, Rudi Zurbuchen, Joshua Ferreri, Emma Huebner, Ron Pauldine, Raquel BartzReviewECMO-facilitated adult non-cardiac surgery: a structured narrative reviewThomas F. O’Shea, Elizabeth J. Bashian, Sarah Y. Park, Rachel Carpenter, Jordan R. H. Hoffman, Michael T. Cain, Markus Kowalsky, Benjamin Abrams, Nathan Clendenen, Nicholas R. Teman, Barbara WilkeyField Grand ChallengeField grand challenge: beyond putting patients asleep: redefining the mission of anesthesiologyAnne-Françoise Rousseau, Thomas Schricker