Timing of delivery in a high-risk obstetric population: a clinical prediction model

Deep Learning Model Using Continuous Skin Temperature Data Predicts Labor Onset

Background

Changes in body temperature anticipate labor onset in numerous mammals, yet this concept has not been explored in humans.

Methods

We evaluated patterns in continuous skin temperature data in 91 pregnant women using a wearable smart ring. Additionally, we collected daily steroid hormone samples leading up to labor in a subset of 28 pregnancies and analyzed relationships among hormones and body temperature trajectory. Finally, we developed a novel autoencoder long-short-term-memory (AE-LSTM) deep learning model to provide a daily estimation of days until labor onset.

Results

Features of temperature change leading up to labor were associated with urinary hormones and labor type. Spontaneous labors exhibited greater estriol to α-pregnanediol ratio, as well as lower body temperature and more stable circadian rhythms compared to pregnancies that did not undergo spontaneous labor. Skin temperature data from 54 pregnancies that underwent spontaneous labor between 34 and 42 weeks of gestation were included in training the AE-LSTM model, and an additional 40 pregnancies that underwent artificial induction of labor or Cesarean without labor were used for further testing. The model was trained only on aggregate 5-minute skin temperature data starting at a gestational age of 240 until labor onset. During cross-validation AE-LSTM average error (true - predicted) dropped below 2 days at 8 days before labor, independent of gestational age. Labor onset windows were calculated from the AE-LSTM output using a probabilistic distribution of model error. For these windows AE-LSTM correctly predicted labor start for 79% of the spontaneous labors within a 4.6-day window at 7 days before true labor, and 7.4-day window at 10 days before true labor.

Conclusion

Continuous skin temperature reflects progression toward labor and hormonal status during pregnancy. Deep learning using continuous temperature may provide clinically valuable tools for pregnancy care.

Predicting labor onset relative to the estimated date of delivery using smart ring physiological data

The transition from pregnancy into parturition is physiologically directed by maternal, fetal and placental tissues. We hypothesize that these processes may be reflected in maternal physiological metrics. We enrolled pregnant participants in the third-trimester (n = 118) to study continuously worn smart ring devices monitoring heart rate, heart rate variability, skin temperature, sleep and physical activity from negative temperature coefficient, 3-D accelerometer and infrared photoplethysmography sensors. Weekly surveys assessed labor symptoms, pain, fatigue and mood. We estimated the association between each metric, gestational age, and the likelihood of a participant's labor beginning prior to (versus after) the clinical estimated delivery date (EDD) of 40.0 weeks with mixed effects regression. A boosted random forest was trained on the physiological metrics to predict pregnancies that naturally passed the EDD versus undergoing onset of labor prior to the EDD. Here we report that many raw sleep, activity, pain, fatigue and labor symptom metrics are correlated with gestational age. As gestational age advances, pregnant individuals have lower resting heart rate 0.357 beats/minute/week, 0.84 higher heart rate variability (milliseconds) and shorter durations of physical activity and sleep. Further, random forest predictions determine pregnancies that would pass the EDD with accuracy of 0.71 (area under the receiver operating curve). Self-reported symptoms of labor correlate with increased gestational age and not with the timing of labor (relative to EDD) or onset of spontaneous labor. The use of maternal smart ring-derived physiological data in the third-trimester may improve prediction of the natural duration of pregnancy relative to the EDD.

A nomogram to optimize the timing of antenatal corticosteroids in threatened preterm delivery

BACKGROUND

Recent observational studies reported a high rate of suboptimal use of antenatal corticosteroids (too anticipated or retrospectively not indicated) for women at risk of preterm delivery despite a recommended use within 7 days before delivery.

OBJECTIVE

This study aimed to elaborate a nomogram aiming at optimizing the timing of administration of antenatal corticosteroids in case of threatened preterm labor, asymptomatic short cervix, or uterine contractions.

STUDY DESIGN

This was an observational retrospective study conducted in a tertiary hospital. All women between 24 and 34 weeks of gestation who received corticosteroids during hospitalization for threatened preterm delivery, asymptomatic short cervix, or uterine contractions requiring tocolysis between 2015 and 2019 were included. Clinical, biological, and sonographic data of women were used to construct logistic regression models for predicting delivery within 7 days. The model was validated on an independent series of women hospitalized in 2020.

RESULTS

Among the 1343 women included in this study, the risk factors independently associated with a delivery within 7 days in multivariate analysis were vaginal bleeding (odds ratio, 14.47; 95% confidence interval, 7.81-26.81; P<.001); need for a second-line tocolysis, such as atosiban (odds ratio, 5.66; 95% confidence interval, 3.39-9.45; P<.001); C-reactive protein level (per 1 mg/L increase; odds ratio, 1.03; 95% confidence interval, 1.02-1.04; P<.001); cervical length (per 1 mm increase; odds ratio, 0.84; 95% confidence interval, 0.82-0.87; P<.001); uterine scar (odds ratio, 2.98; 95% confidence interval, 1.33-6.65; P=.008), and gestational age at admission (per week of amenorrhea increase; odds ratio, 1.10; 95% confidence interval, 1.00-1.20; P=.041). Based on these results, a nomogram was developed that, in retrospect, would have allowed physicians to avoid or delay antenatal corticosteroids in 57% of cases in our population. The discrimination of the predictive model was good when applied to the validation set of 232 women hospitalized in 2020. It would have enabled physicians to avoid or delay antenatal corticosteroids in 52% of cases.

CONCLUSION

This study developed a simple use, accurate prognostic score to identify women at risk of delivery within 7 days in cases of threatened preterm delivery, asymptomatic short cervix, or uterine contractions and thereby optimized the use of antenatal corticosteroids.

Overly optimistic prediction results on imbalanced data: a case study of flaws and benefits when applying over-sampling

Information extracted from electrohysterography recordings could potentially prove to be an interesting additional source of information to estimate the risk on preterm birth. Recently, a large number of studies have reported near-perfect results to distinguish between recordings of patients that will deliver term or preterm using a public resource, called the Term/Preterm Electrohysterogram database. However, we argue that these results are overly optimistic due to a methodological flaw being made. In this work, we focus on one specific type of methodological flaw: applying over-sampling before partitioning the data into mutually exclusive training and testing sets. We show how this causes the results to be biased using two artificial datasets and reproduce results of studies in which this flaw was identified. Moreover, we evaluate the actual impact of over-sampling on predictive performance, when applied prior to data partitioning, using the same methodologies of related studies, to provide a realistic view of these methodologies' generalization capabilities. We make our research reproducible by providing all the code under an open license.

Clinical information extraction for preterm birth risk prediction

This paper contributes to the pursuit of leveraging unstructured medical notes to structured clinical decision making. In particular, we present a pipeline for clinical information extraction from medical notes related to preterm birth, and discuss the main challenges as well as its potential for clinical practice. A large collection of medical notes, created by staff during hospitalizations of patients who were at risk of delivering preterm, was gathered and analyzed. Based on an annotated collection of notes, we trained and evaluated information extraction components to discover clinical entities such as symptoms, events, anatomical sites and procedures, as well as attributes linked to these clinical entities. In a retrospective study, we show that these are highly informative for clinical decision support models that are trained to predict whether delivery is likely to occur within specific time windows, in combination with structured information from electronic health records.

Patterns of antenatal corticosteroid administration in a cohort of women with diabetes in pregnancy

Antenatal corticosteroids administered to the mother prior to birth decrease the risk of mortality and major morbidity in infants born at less than 35 weeks’ gestation. However, the evidence relating to women with diabetes in pregnancy is limited. Clinical guidelines for antenatal corticosteroid administration recommend that women with diabetes in pregnancy are treated in the same way as women without diabetes, but there are no recent descriptions of whether contemporary practice complies with this guidance. This study is a retrospective review of antenatal corticosteroid administration at a New Zealand tertiary hospital in women with diabetes in pregnancy. We found that in this cohort, for both an initial course at less than 35 weeks’ gestation and repeat courses at less than 33 weeks’, the administration of antenatal corticosteroid to women with diabetes in pregnancy is largely consistent with current Australian and New Zealand recommendations. However, almost 25% of women received their last dose of antenatal corticosteroid at or beyond the latest recommended gestation of 35 weeks’ gestation. Pre-existing diabetes and planned caesarean section were independently associated with an increased rate of antenatal corticosteroid administration. We conclude that diabetes in pregnancy does not appear to be a deterrent to antenatal corticosteroid administration. The high rates of administration at gestations beyond recommendations, despite the lack of evidence of benefit in this group of women, highlights the need for further research into the risks and benefits of antenatal corticosteroid administration to women with diabetes in pregnancy, particularly in the late preterm and early term periods.

Association between maternal serious mental illness and adverse birth outcomes

OBJECTIVE

To evaluate the contribution of serious mental illness (SMI) and specific risk factors (comorbidities and substance use) to the risk of adverse birth outcomes.

STUDY DESIGN

This cross-sectional study uses maternal delivery records in the Healthcare Cost and Utilization Project Nationwide/National Inpatient Sample (HCUP-NIS) to estimate risk factor prevalence and relative risk of adverse birth outcomes (e.g., preeclampsia, preterm birth, and fetal distress) in women with SMI.

RESULTS

The relative risk of adverse gestational (1.15, 95% CI: 1.13-1.17), obstetric (1.07, 1.06-1.08), and fetal (1.24, 1.21-1.26) outcomes is increased for women with SMI. After adjusting for risk factors, the risk is significantly reduced but remains elevated for all three adverse outcome categories (gestational: 1.08, 1.06-1.09; obstetric: 1.03, 1.02-1.05; fetal: 1.12, 1.09-1.14).

CONCLUSIONS

Maternal serious mental illness is independently associated with increased risk for adverse birth outcomes. However, approximately half of the excess risk is attributable to comorbidities and substance use.