Bayesian synthetic likelihood (BSL) is an established method for performing approximate Bayesian inference when the likelihood function is intractable. In synthetic likelihood methods, the likelihood function is approximated parametrically via model simulations, and then standard likelihood-based techniques are used to perform inference. The Gaussian synthetic likelihood estimator has become ubiquitous in BSL literature, primarily for its simplicity and ease of implementation. However, it is often too restrictive and may lead to poor posterior approximations. Recently, a more flexible semi-parametric Bayesian synthetic likelihood (semiBSL) estimator has been introduced, which is significantly more robust to irregularly distributed summary statistics. A number of extensions to semiBSL are proposed. First, even more flexible estimators of the marginal distributions are considered, using transformation kernel density estimation. Second, whitening semiBSL (wsemiBSL) is proposed – a method to significantly improve the computational efficiency of semiBSL. wsemiBSL uses an approximate whitening transformation to decorrelate summary statistics at each algorithm iteration. The methods developed herein significantly improve the versatility and efficiency of BSL algorithms.

INTRODUCTION:Maternal obesity is a significant risk factor for poor pregnancy outcomes. Obesity is linked to abnormalities in placental tissue at term. The purpose of this study was to correlate changes in placental stiffness, measured via ultrasound elastography, with maternal pre-pregnancy body mass index and gestational weight gain. METHODS:Body Mass Index and gestation weight gain data was collected from 238 women. Elastography measurements were obtained via ultrasound at 24-, 28- and 36-weeks' gestation. An analysis using a linear mixed regression model assessed for the statistical significance of pre-pregnancy BMI, pregnancy weight gain and placental SWV (Shear Wave Velocity). RESULTS:Pre-pregnancy weight status has a significant impact on placental tissue stiffness detectable via ultrasound elastography. Placental tissue stiffness was highest in obese women, followed by overweight women. Obese women, on average, had a MeanSWV 0.11 m/s (95% CI (0.061-0.15) m/s, p < 0.001) above the healthy group throughout the 3rd trimester. Weight gain during pregnancy had a small impact on placental stiffness at the end of pregnancy. MeanSWV was 0.06 m/s (95% CI (0.03-0.10) m/s, p < 0.001) higher in the excessive weight gain group. DISCUSSION:Structural changes of the placenta detected via ultrasound elastography techniques are not exclusive to placental dysfunction conditions (pre-eclampsia and growth restriction) but are also associated with maternal obesity.

INTRODUCTION:Ultrasound elastography shows diagnostic promise via the non-invasive determination of placental elastic properties. A limitation is a potential for inadequate measurements from posterior placentae. This study aimed to analyse placental position's influence on measures of shear wave elastography (SWV). METHODS:SWV elastography measurements were obtained via ultrasound at 24, 28 and 36 weeks gestation from 238 pregnancies. . The placental position was labelled as either anterior, posterior or fundal/lateral. Average SWV measurements (m/s) and the corresponding standard deviations (SD) were used for data analysis. RESULTS:There was a statistically significant difference between SWV recorded from anterior (1.33 ± 0.19)m/s and posterior (1.39 ± 0.18)m/s placentae (p < 0.001). However, the average sampling depth between these groups was significantly different (3.98 cm vs. 5.38 cm, p < 0.001). There was no statistically significant difference between SWV when measurements were compared at similar depths, regardless of placental location. The addition of placental position to a previously developed mixed-effects model confirmed placental position did not result in improved SWV measurements. In this model, sampling depth remained the best predictor for SWV. CONCLUSIONS:This study showed that placental position does not influence the accuracy or reliability of SWV.

Likelihood-free methods are an established approach for performing approximate Bayesian inference for models with intractable likelihood functions. However, they can be computationally demanding. Bayesian synthetic likelihood (BSL) is a popular such method that approximates the likelihood function of the summary statistic with a known, tractable distribution—typically Gaussian—and then performs statistical inference using standard likelihood-based techniques. However, as the number of summary statistics grows, the number of model simulations required to accurately estimate the covariance matrix for this likelihood rapidly increases. This poses a significant challenge for the application of BSL, especially in cases where model simulation is expensive. In this article, we propose whitening BSL (wBSL)—an efficient BSL method that uses approximate whitening transformations to decorrelate the summary statistics at each algorithm iteration. We show empirically that this can reduce the number of model simulations required to implement BSL by more than an order of magnitude, without much loss of accuracy. We explore a range of whitening procedures and demonstrate the performance of wBSL on a range of simulated and real modeling scenarios from ecology and biology. Supplementary materials for this article are available online.

Introduction: Research into the role of ultrasound elastography to assess compromised placental tissue is ongoing. There is particular interest in evaluating its potential in the investigation of changes associated with uteroplacental dysfunction. To date, there is limited data on how different maternal and fetal considerations, such as advancing gestational age, amniotic fluid Index (AFI) and maternal body mass index (BMI) may influence shear wave velocity (SWV) measurements. This study aimed to evaluate longitudinal changes in SWV throughout gestation and model these changes with other developing fetal and maternal physiological and biological characteristics. Methods: The study utilised 238 singleton pregnancies and collected longitudinal data at repeated intervals in the 3rd trimester representing 629 individual data points. Linear mixed model regression analysis was used to identify significant predictors for SWV. Results: From a total of ten variables selected for modelling, only gestational age, AFI, BMI, and sample depth were found to be significant predictors of placental SWV, and gestational age and AFI were found to have only a minimal impact on SWV. Discussion: Sophisticated statistical modelling demonstrates that many of the expected maternal and fetal changes in the 3rd trimester have no or minimal impact on placental SWV. Understanding which factors influence placental SWV is essential to ascertain the technique's utility in managing pregnancies complicated by placental dysfunction in the future.