Machine Learning for the Prediction of Surgical Morbidity in Placenta Accreta Spectrum.

OBJECTIVE: We sought to create a machine learning (ML) model to identify variables that would aid in the prediction of surgical morbidity in cases of placenta accreta spectrum (PAS). STUDY DESIGN: A multicenter analysis including all cases of PAS identified by pathology specimen confirmation, across five tertiary care perinatal centers in New York City from 2013 to 2022. We developed models to predict operative morbidity using 213 variables including demographics, obstetrical information, and limited prenatal imaging findings detailing placental location. Our primary outcome was prediction of a surgical morbidity composite defined as including any of the following: blood loss (>1,500 mL), transfusion, intensive care unit admission, vasopressor use, mechanical ventilation/intubation, and organ injury. A nested, stratified, cross-validation approach was used to tune model hyperparameters and estimate generalizability. Gradient boosted tree classifier models incorporated preprocessing steps of standard scaling for numerical variables and one-hot encoding for categorical variables. Model performance was evaluated using area under the receiver operating characteristic curve (AUC), positive and negative predictive values (PPV, NPV), and F1 score. Variable importance ranking was also determined. RESULTS: Among 401 PAS cases, 326 (81%) underwent hysterectomy. Of the 401 cases of PAS, 309 (77%) had at least one event defined as surgical morbidity. Our predictive model had an AUC of 0.79 (95% confidence interval: 0.69, 0.89), PPV 0.79, NPV 0.76, and F1 score of 0.88. The variables most predictive of surgical morbidity were completion of a hysterectomy, prepregnancy body mass index (BMI), absence of a second trimester ultrasound, socioeconomic status zip code, BMI at delivery, number of prenatal visits, and delivery time of day. CONCLUSION: By identifying social and obstetrical characteristics that increase patients' risk, ML models are useful in predicting PAS-related surgical morbidity. Utilizing ML could serve as a foundation for risk and complexity stratification in cases of PAS to optimize surgical planning. KEY POINTS: · ML models are useful models are useful in predicting PAS-related surgical morbidity.. · Optimal management for PAS remains unclear.. · Utilizing ML can serve as a foundation for risk and complexity stratification in cases of PAS..

Use of natural language processing to uncover racial bias in obstetrical documentation.

Natural Language Processing (NLP), a form of Artificial Intelligence, allows free-text based clinical documentation to be integrated in ways that facilitate data analysis, data interpretation and formation of individualized medical and obstetrical care. In this cross-sectional study, we identified all births during the study period carrying the radiology-confirmed diagnosis of fibroid uterus in pregnancy (defined as size of largest diameter of >5 cm) by using an NLP platform and compared it to non-NLP derived data using ICD10 codes of the same diagnosis. We then compared the two sets of data and stratified documentation gaps by race. Using fibroid uterus in pregnancy as a marker, we found that Black patients were more likely to have the diagnosis entered late into the patient's chart or had missing documentation of the diagnosis. With appropriate algorithm definitions, cross referencing and thorough validation steps, NLP can contribute to identifying areas of documentation gaps and improve quality of care.

Investigation of health inequities in maternal and neonatal outcomes of patients with placenta accreta spectrum: a multicenter study.

BACKGROUND: Placenta accreta spectrum is associated with significant maternal and neonatal morbidity and mortality. There is limited established data on healthcare inequities in the outcomes of patients with placenta accreta spectrum. OBJECTIVE: This study aimed to investigate health inequities in maternal and neonatal outcomes of pregnancies with placenta accreta spectrum. STUDY DESIGN: This multicentered retrospective cohort study included patients with a histopathological diagnosis of placenta accreta spectrum at 4 regional perinatal centers between January 1, 2013, and June 30, 2022. Maternal race and ethnicity were categorized as either Hispanic, non-Hispanic Black, non-Hispanic White, or Asian or Pacific Islander. The primary outcome was a composite adverse maternal outcome: transfusion of ≥4 units of packed red blood cells, vasopressor use, mechanical ventilation, bowel or bladder injury, or mortality. The secondary outcomes were a composite adverse neonatal outcome (Apgar score of <7 at 1 minute, morbidity, or mortality), gestational age at placenta accreta spectrum diagnosis, and planned delivery by a multidisciplinary team. Multivariable logistic regression was used to estimate the associations of race and ethnicity with maternal and neonatal outcomes. RESULTS: A total of 408 pregnancies with placenta accreta spectrum were included. In 218 patients (53.0%), the diagnosis of placenta accreta spectrum was made antenatally. Patients predominantly self-identified as non-Hispanic White (31.6%) or non-Hispanic Black (24.5%). After adjusting for institution, age, body mass index, income, and parity, there was no difference in composite adverse maternal outcomes among the racial and ethnic groups. Similarly, adverse neonatal outcomes, gestational age at prenatal diagnosis, rate of planned delivery by a multidisciplinary team, and cesarean hysterectomy were similar among groups. CONCLUSION: In our multicentered placenta accreta spectrum cohort, race and ethnicity were not associated with inequities in composite maternal or neonatal morbidity, timing of diagnosis, or planned multidisciplinary care. This study hypothesized that a comparable incidence of individual risk factors for perinatal morbidity and geographic proximity reduces potential inequities that may exist in a larger population.

The case for making the first-trimester anatomical survey a standard of care post Dobbs.

Currently, 11- to 14-week-detailed anatomic surveys are generally reserved for at-risk populations because of the lower incidence of major fetal anomalies in low-risk populations. Until recently, such standard reflects, in part, the fact that pregnant persons retain the option of abortion even if the initial anatomy scan was in the second trimester of pregnancy. However, on June 24, 2022, the US Supreme Court overturned Roe, and many states subsequently lowered the gestational age at which abortions can legally be performed. Here, we argue for a reconsideration of limitations on first-trimester scans to preserve pregnant persons' reproductive options, particularly in those states that have imposed laws limiting access to abortion. Moreover, we acknowledge and discuss some of the challenges that will be associated with this approach.

Use of Machine Learning to Identify Clinical Variables in Pregnant and Non-Pregnant Women with SARS-CoV-2 Infection.

OBJECTIVE: The aim of the study is to identify the important clinical variables found in both pregnant and non-pregnant women who tested positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, using an artificial intelligence (AI) platform. METHODS: This was a retrospective cohort study of all women between the ages of 18 to 45, who were admitted to Maimonides Medical Center between March 10, 2020 and December 20, 2021. Patients were included if they had nasopharyngeal PCR swab positive for SARS-CoV-2. Safe People Artificial Intelligence (SPAI) platform, developed by Gynisus, Inc., was used to identify key clinical variables predicting a positive test in pregnant and non-pregnant women. A list of mathematically important clinical variables was generated for both non-pregnant and pregnant women. RESULTS: Positive results were obtained in 1,935 non-pregnant women and 1,909 non-pregnant women tested negative for SARS-CoV-2 infection. Among pregnant women, 280 tested positive, and 1,000 tested negative. The most important clinical variable to predict a positive swab result in non-pregnant women was age, while elevated D-dimer levels and presence of an abnormal fetal heart rate pattern were the most important clinical variable in pregnant women to predict a positive test. CONCLUSION: In an attempt to better understand the natural history of the SARS-CoV-2 infection we present a side-by-side analysis of clinical variables found in pregnant and non-pregnant women who tested positive for COVID-19. These clinical variables can help stratify and highlight those at risk for SARS-CoV-2 infection and shed light on the individual patient risk for testing positive.

A Novel and Precise Profiling Tool to Predict Gestational Diabetes.

BACKGROUND: There is a trend in healthcare for developing models for predictions of disease to enable early intervention and improve outcome. INSTRUMENT: We present the use of artificial intelligence algorithms that were developed by Gynisus Ltd. using mathematical algorithms. EXPERIENCE: Data were retrospectively collected on pregnant women that delivered at a single institution. Hundreds of parameters were collected and found to have different importance and correlation with the likelihood to develop gestational diabetes mellitus (GDM). We highlight 3 of 29 specific parameters that were important in pregestation and in early pregnancy, which have not been previously correlated with GDM. CONCLUSION: This predictive tool identified parameters that are not currently being used as predictors in GDM, even before pregnancy. This tool opens the possibility of intervening on patients identified at risk for GDM and its complications. Future prospective studies are needed.

A Novel Partogram for Stages 1 and 2 of Labor Based on Fetal Head Station Measured by Ultrasound: A Prospective Multicenter Cohort Study.

OBJECTIVE: This study was aimed to describe continuous labor curves, including second stage, based on fetal head station. STUDY DESIGN: We performed a prospective multicenter cohort study. The inclusion criteria were women with singleton uncomplicated cephalic term pregnancies in labor, who delivered vaginally. We used a device that combines ultrasound imaging with position-tracking technology to monitor the head station noninvasively throughout labor. We collected data on demographics, labor parameters, and delivery and neonatal outcomes. RESULTS: A total of 613 women delivered vaginally, 327 (53.3%) were nulliparous, while 286 (46.7%) were multiparous. Time to delivery (TTD) diminished progressively with descent of the fetal head. When the head is engaged, the labor curve of multiparous women demonstrated a more prominent downward shift in curve as compared with nulliparous women. When comparing multipara and nullipara at engagement level, the median TTD was 1 and 1.62 hours, respectively. In 95% of women with unengaged head during the second stage, TTD of nulliparous and multiparous women were less than 3.8 and 3 hours, respectively. CONCLUSION: While current labor curves end at full dilatation, the described curves were developed throughout stages 1 and 2 of labor. The TTD, according to the station curves, shows an acceleration of labor, once passed the engagement level, especially in multiparous women.

Caring for Pregnant Patients with COVID-19: Practical Tips Getting from Policy to Practice.

Novel coronavirus disease 2019 (COVID-19) is a pandemic with most American cases in New York. As an institution residing in a high-prevalence zip code, with over 8,000 births annually, we have cared for over 80 COVID-19-infected pregnant women, and have encountered many challenges in applying new national standards for care. In this article, we review how to change outpatient and inpatient practices, develop, and disseminate new hospital protocols, and we highlight the psychosocial challenges for pregnant patients and their providers. KEY POINTS: · Novel coronavirus disease 2019 (COVID-19) information rapidly changes.. · Multidisciplinary communication is key.. · This study addresses psychosocial challenges..

The Relationship between Status at Presentation and Outcomes among Pregnant Women with COVID-19.

OBJECTIVE: This study was aimed to compare maternal and pregnancy outcomes of symptomatic and asymptomatic pregnant women with novel coronavirus disease 2019 (COVID-19). STUDY DESIGN: This is a retrospective cohort study of pregnant women with COVID-19. Pregnant women were divided into two groups based on status at admission, symptomatic or asymptomatic. All testing was done by nasopharyngeal swab using polymerase chain reaction (PCR) for severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2). Initially, nasopharyngeal testing was performed only on women with a positive screen (symptoms or exposure) but subsequently, testing was universally performed on all women admitted to labor and delivery. Chi-square and Wilcoxon's rank-sum tests were used to compare outcomes between groups. RESULTS: Eighty-one patients were tested because of a positive screen (symptoms [n = 60] or exposure only [n = 21]) and 75 patients were universally tested (all asymptomatic). In total, there were 46 symptomatic women and 22 asymptomatic women (tested based on exposure only [n = 12] or as part of universal screening [n = 10]) with confirmed COVID-19. Of symptomatic women (n = 46), 27.3% had preterm delivery and 26.1% needed respiratory support while none of the asymptomatic women (n = 22) had preterm delivery or need of respiratory support (p = 0.007 and 0.01, respectively). CONCLUSION: Pregnant women who presented with COVID19-related symptoms and subsequently tested positive for COVID-19 have a higher rate of preterm delivery and need for respiratory support than asymptomatic pregnant women. It is important to be particularly rigorous in caring for COVID-19 infected pregnant women who present with symptoms. KEY POINTS: · Respiratory support is often needed for women who present with symptoms.. · Low rate of severe disease in women who present without symptoms.. · There were no neonatal infections on day 0 of life..

The effect of maternal position on fetal middle cerebral artery Doppler indices and its association with adverse perinatal outcomes: a pilot study.

Objective The aim of this study was to compare position-related changes in fetal middle cerebral artery (MCA) Doppler pulsatility indices (PI). Methods A prospective study of 41 women with conditions associated with placental-pathology (chronic hypertension, pregestational diabetes, and abnormal analytes) and 34 women without those conditions was carried out. Fetal MCA Doppler velocity flow waveforms were obtained in maternal supine and left lateral decubitus positions. MCA PI Δ was calculated by subtracting the PI in the supine position from the PI in the left lateral position. Secondary outcomes included a composite of adverse perinatal outcomes (fetal growth restriction, oligohydramnios, and preeclampsia). χ2 and Student t-tests and repeated-measures analysis of variance were used. Results MCA PI Δ was significantly less for high-risk pregnant women ([P = 0.03]: high risk, left lateral PI, 1.90 ± 0.45 vs. supine PI, 1.88 ± 0.46 [Δ = 0.02]; low risk, left lateral PI, 1.90 ± 0.525 vs. supine PI, 1.68 ± 0.40 [Δ = 0.22]). MCA PI Δ was not significantly different between women who had a composite adverse outcome and women who did not have a composite adverse outcome (P = 0.843). Conclusion Our preliminary study highlights differences in position-related changes in fetal MCA PI between high-risk and low-risk pregnancies. These differences could reflect an attenuated ability of women with certain risk factors to respond to physiologic stress.

Predicting Preeclampsia with Noninvasive Measures of Endothelial Dysfunction: A Pilot Study.

Objective This study evaluates the assessment of endothelial function and its prediction for preeclampsia among women with high-risk factors. Study Design A prospective cohort study of 107 pregnant women at 20 weeks or greater gestation with risk factors for developing preeclampsia. Endothelial dysfunction was assessed using peripheral arterial tonometry by generating a reactive hyperemia index (RHI) score. An index score of <1.67 was defined as endothelial dysfunction. The primary outcome was preeclampsia. Logistic regression was used to predict preeclampsia from RHI scores, body mass index, gestational age at RHI evaluation, history of preeclampsia, history of pregestational diabetes mellitus, chronic hypertension, and fetal number. A receiver operating characteristic plot was constructed to predict preeclampsia from the RHI score. Results Among 107 women, 99 had interpretable RHI scores. Among those with an abnormal RHI ( n = 61), 17 (28%) developed preeclampsia. Among women with a normal score ( n = 38), six (16%) developed preeclampsia ( p = 0.166). After logistic regression, there was no significant association. A receiver operating characteristic plot also revealed no association between RHI score and preeclampsia. Conclusion An abnormal RHI score using peripheral arterial tonometry indicating endothelial dysfunction was not predictive of developing preeclampsia in this cohort. Future studies are needed to further evaluate this relationship.

Same-day confirmation of intrauterine pregnancy failure in women with first- and early second-trimester bleeding.

OBJECTIVE: To determine if alpha-fetoprotein (AFP) concentration in vaginal blood, in the setting of dissolved fetal tissue, is significantly higher than its concentration in the maternal serum. DESIGN: A prospective cohort study. SETTING: Medical center. PATIENT(S): Four groups of women were evaluated: 1) with missed/incomplete miscarriage with vaginal bleeding; 2) with threatened miscarriage; 3) with vaginal bleeding during cerclage placement; and 4) undergoing dilation and curettage (D&C). INTERVENTIONS(S): None. MAIN OUTCOME MEASURE(S): In each patient, AFP concentration in the vaginal blood or in the liquid component of the evacuated products of conception (POC; D&C group) was compared with the AFP concentration in the maternal serum. RESULT(S): The median (range) concentration ratios of AFP in vaginal blood (or POC) to AFP in maternal serum were 24.5 (5.1-8,620) and 957 (4.6-24,216) for the missed/incomplete (n = 30) and the D&C (n = 22) groups, respectively, whereas they were only 1.2 (0.4-13.4) and 1.01 (0.7-1.5) for the threatened miscarriage (n = 15) and cerclage (n = 9) groups, respectively. Receiver operating characteristic (ROC) analysis demonstrated 100% sensitivity and 86.7% specificity for the detection of the passage of fetal tissue (ratio 4.3, area under the ROC curve 0.96). CONCLUSION(S): Higher concentrations of AFP in vaginal blood than in maternal serum may indicate the presence of dissolved fetal tissue (i.e., confirming a failed pregnancy).

Sonographic Measurement of Cervical Volume in Pregnant Women at High Risk of Preterm Birth Using a Geometric Formula for a Frustum Versus 3-Dimensional Automated Virtual Organ Computer-Aided Analysis.

OBJECTIVES: To compare cervical volume measurements by 3-dimensional (3D) sonography using Virtual Organ computer-aided analysis (VOCAL; GE Healthcare, Milwaukee, WI) versus a manual method using a geometric formula for a frustum. METHODS: We included 142 asymptomatic pregnant women at 16 to 24 weeks gestation at high risk for preterm birth. With a Voluson 730 Expert system (GE Healthcare), they underwent 2-dimensional (2D) transvaginal sonographic cervical length measurements and 3D cervical volume acquisition. The stored volumes were processed by VOCAL on a surface tablet. Cervical volume was manually calculated from the 2D images by using the formula V = 1/3 × π × h × (r12 + r22 + r1 × r2), where V represents cervical volume; π was approximated as 3.14159; h, cervical length; r1, radius at the internal os; and r2, radius at the external os. RESULTS: Cervical volume was lower when obtained manually than by VOCAL, with a coefficient of variation of 30%, a mean difference of 10.1 ± 14.9 cm3 (P < .0001), and a poor interclass correlation coefficient of 0.62 (95% confidence interval [CI], 0.31 to 0.78). Both methods had good reproducibility; however, VOCAL had wider limits of agreement. A positive correlation was found between both methods (r = 0.63; P < .0001). No correlation was found between cervical length by 2D transvaginal ultrasound and cervical volume by the VOCAL technique (r = 0.06; 95% CI, -0.10 to 0.22) or cervical volume by the manual method (r = 0.2; 95% CI, 0.08 to 0.39). CONCLUSIONS: The cervix represents a frustum (truncated cone, r1 is not equal to r2) in shape rather than a cylinder. Both methods are reproducible; VOCAL is less reliable but provides higher values of cervical volume.

Fetal proximal humeral epiphysis as an indicator of term gestation in different ethnic groups.

OBJECTIVES: Accurate pregnancy dating is critical for appropriate clinical management. Our aim was to determine the time of appearance of proximal humeral epiphysis (PHE), consistency of its appearance among ethnic groups and whether 3D imaging helps with its visualization. METHODS: A cross-sectional study was done on 360 patients with 563 scans in different ethnic groups between August 2013 and July 2015. Inclusion criteria were singleton pregnancies (34-40+ weeks of gestation), well dated by <20 weeks sonogram. RESULTS: PHE was not seen at 34 (n = 44) or 35 weeks (n = 36) and was present at gestational ages 36 (n = 3), 37 (n = 126), 38 (n = 96), 39 (n = 100) and 40 weeks (n = 28) in 2%, 12%, 51%, 75% and 100%, respectively. PHE was seen in 20 of 50 (60%) African-Americans, 22 of 61 (64%) south Asians, 41 of 72 (57%) Caucasians, 45 of 86 (48%) Hispanics and 41 of 80 (49%) Asians. CONCLUSION: Appearance of PHE did increase with gestational age, prior to 40 weeks, it was not uniformly present and was seen as early as 36 weeks independent of ethnic group.

A Sanitary Pad for Self-Assessment of Rupture of Membranes.

BACKGROUND: Multistep immunoassay kits for the diagnosis of rupture of membranes are relatively complex and are not designed to be used by pregnant women themselves. These kits require procedural steps of specimen extraction and preparation. We evaluated the ability of a sanitary pad containing a qualitative immunoassay for alpha-fetoprotein (AFP) to serve as a one-step self-test to detect amniotic fluid leakage. TECHNIQUE: Four sets of pads were evaluated. The pads in the study set were worn by 288 pregnant women with confirmed rupture of membranes. Three controls were evaluated: 1) pads worn by 93 pregnant women with intact membranes, 2) additional pads instilled with urine specimens obtained from the 381 women described previously (study set plus control set 1), and 3) pads instilled with semen collected from 40 men. EXPERIENCE: All 288 pads that absorbed amniotic fluid had positive results. Approximately half of the pads absorbed with normal vaginal discharge had a sufficient amount to yield valid results, which were all negative. All 381 pads with instilled urine and all 40 pads with instilled semen had negative results. CONCLUSION: An immunoassay for AFP, embedded in a pad, appears to be a feasible and reproducible self-test for the detection of rupture of membranes.

Uterine tachysystole in spontaneous labor at term.

OBJECTIVES: The objective of this study is to determine the incidence of uterine tachysystole and its association with spontaneous labor at term. METHODS: A retrospective cohort study of 8008 women in spontaneous labor (without prostaglandins or oxytocin). Fetal heart tracings and uterine activity were recorded every 15 min. PRIMARY OUTCOME: occurrence of tachysystole (> 5 uterine contractions /10 min over 30 min periods). SECONDARY OUTCOMES: non-reassuring fetal heart tracings (NRFHT), NICU admissions, and cesarean deliveries. RESULTS: About 890 patients (11.1 %) had at least one episode of tachysystole. Non-whites have higher incidence of uterine tachysystole; adjusted odds ratio (aOR) was 1.66 for Hispanics (95% CI 1.28-2.05), 1.58 for African Americans (95% CI 1.05-2.38), and 1.51 for Asians (95% CI = 1.13-2.0). The use of epidural analgesia was higher in the tachysystole group (62.2% versus 40.9%, aOR 1.89, CI 1.58-2.26; p < 0.001). Tachysystole was more frequent among nulliparous women and in women carrying higher weight fetuses. Oligohydramnios (aOR 1.62, CI 0.70-3.72; p < 0.004), and NRFHT were more common in the tachysystole group (4.2% versus 2.5%, p = 0.002). Newborns in the tachysystole group were two times more likely to be admitted to NICU (30 /890 [3.4%] versus 122 /7118 [1.7%], OR = 2, p=0.001). There was no difference in the frequency of meconium-stained amniotic fluid or Apgar scores <7 at 5 min. CONCLUSION: Uterine tachysystole occurs in more than 10% of spontaneous labors and is associated with NRFHR, increased rate of caesarean deliveries and NICU admissions. It is not associated with low Apgar scores or meconium-stained amniotic fluid.

Carcinoembryonic antigen as a biomarker for meconium-stained amniotic fluid.

OBJECTIVE: To assess whether elevated carcinoembryonic antigen (CEA) concentration in amniotic fluid can indicate meconium-stained amniotic fluid (MSAF). METHODS: In a prospective cohort study, women with a term singleton pregnancy who were in labor but had intact membranes were recruited at a center in Israel over a 5-month period in 2013. Only women who subsequently underwent artificial rupture of membranes following a clear medical indication were included. Samples of amniotic fluid, urine, and serum were collected. Amniotic fluid was examined by sight and classified as clear, MSAF, or undetermined. CEA concentration in the samples was measured. RESULTS: Among 81 participants, 45 had clear amniotic fluid, 28 had MSAF, and eight had undetermined amniotic fluid. Mean CEA concentration was more than 10 times higher in MSAF (2658 µg/L, standard error 250) than in clear amniotic fluid (238 µg/L, standard error 29; P<0.001). Receiver operating characteristic curve analysis demonstrated a sensitivity of 96% and a specificity of 100% for distinguishing MSAF from clear amniotic fluid at a CEA cutoff of 799.2 µg/L. CEA concentrations in urine and serum were all within the normal range (≤5 µg/L), irrespective of amniotic fluid status. CONCLUSION: High CEA concentrations in amniotic fluid can assist in the diagnosis of MSAF. These findings could provide the basis for a bedside test to detect MSAF following rupture of membranes.

Acute presentation of gestational diabetes insipidus with pre-eclampsia complicated by cerebral vasoconstriction: a case report and review of the published work.

Gestational diabetes insipidus (GDI) is a rare, self-limited complication of pregnancy. As it is related to excess placental vasopressinase enzyme activity, which is metabolized in the liver, GDI is more common in pregnancies complicated by conditions associated with liver dysfunction. We present a case of a 41-year-old woman at 38 weeks' gestation who presented with pre-eclampsia with severe features, including impaired liver function and renal insufficiency. Following cesarean section she was diagnosed with GDI, which was further complicated by cerebral vasoconstriction as demonstrated by magnetic resonance angiography. This case raises the possibility that cerebral vasoconstriction may be related to the cause of GDI. A high index of suspicion of GDI should be maintained in patients who present with typical signs and symptoms, especially in the setting of pregnancy complications associated with liver dysfunction.