Tommy's

Miscarriage, the loss of a pregnancy before viability, can be sporadic or recurrent. Emerging evidence links miscarriage to specific microbiota compositions within the female reproductive tract (FRT). This systematic review aims to synthesise evidence on the association between sporadic and recurrent miscarriage and FRT microbiota composition, as assessed using metataxonomic profiling approaches. A systematic analysis of the 43 included studies, sampling the vaginal, cervical and endometrial microbiota supported an association between reduced Lactobacillus abundance and miscarriage, making it a potential target for therapeutic intervention. However, consistent changes in alpha and beta diversity were not observed and there was a lack of reproducibility for other compositional changes. This review also highlighted concerns about the significant bias introduced due to methodological variations and emphasises the need for future standardisation of microbial sampling, sequencing, and reporting to allow accurate comparison of results and to reduce research waste.

Abstract Study question Can an image recognition model, utilising automated deep learning, be trained to diagnose and classify the severity of adenomyosis on ultrasound? Summary answer The automated deep learning model developed in this study can accurately diagnose and classify the severity of adenomyosis from ultrasound images. What is known already Adenomyosis is prevalent across all stages of life, starting from adolescence to perimenopause, however it remains underdiagnosed. The diagnosis and classification of adenomyosis are essential for correlating with symptom severity and counselling about disease prognosis, management options and long-term impact. Diagnosing and classifying the severity is a time-consuming and complex process that depends on the operator’s experience. Machine learning (ML) has an enormous potential to assist healthcare professionals with image classification tasks. It can directly process and automatically learn mid to high-level abstract features acquired from images using a deep architecture model without requiring the manual definition of features. Study design, size, duration This observational cohort study utilised ultrasound images from four assisted conception units in the United Kingdom from February 2023 to February 2024. Labelled images were used as input for training and testing of the model. The complete dataset was manually split into 9:1 ratio; 90% of all images were used as training dataset (80% for training and 10% for validation) and 10% as test dataset. Ethics approval was obtained from the Institutional Review Board (IRB). Participants/materials, setting, methods An automated ML tool (Vertex AI Vision) was trained and tested to diagnose and classify adenomyosis severity. Two investigators, experienced in gynaecological ultrasound scanning, labelled two and three-dimensional images into four categories: normal uterus, mild, moderate, and severe adenomyotic uterus, as per standardised ultrasound criteria. The labelled images were formatted and uploaded on Google Cloud to train the neural network based on pattern recognition. The model’s performance was determined for an internal test dataset. Main results and the role of chance A total of 3,500 ultrasound images were obtained from 3,194 patients, and the internal test dataset consisted of 350 ultrasound images from 292 patients. An overall and per-category score threshold of 0.5 was adopted to provide a balance between reasonable precision and recall during the evaluation. The investigator’s categorization was used as the reference standard to evaluate the performance of the automated machine-learning model. The model’s overall accuracy for diagnosing and classifying adenomyosis, as determined by the area under the precision-recall curve, was 80.24% across all score thresholds. The overall precision was 83.00%, and recall was 65.75% at a confidence threshold of 0.5. The precision and recall were 91% and 78% for normal uterus, 67% and 53% for mild adenomyosis, 86% and 68% for moderate adenomyosis, and 88% and 64% for severe adenomyosis. The algorithm achieved a high accuracy for the internal dataset. The time taken by a trained healthcare professional to label and process 3,500 images into 4 categories was 70 hours. The time taken by an untrained automated ML tool to get trained, process, and label the images into 4 categories was 4 hours, saving 66 hours per healthcare professional. Limitations, reasons for caution Due to the inherent nature of non-invasive investigations for the population attending the assisted conception unit, histological confirmation of adenomyosis was not possible. In addition, the model’s performance has been tested using an internal test dataset and needs further validation using an external test dataset from a different population. Wider implications of the findings This novel study leverages the power of artificial intelligence to develop an automated machine learning model for diagnosing and classifying adenomyosis from ultrasound scans. This tool can potentially enhance the diagnostic capacity and efficiency of healthcare professionals and empower women by providing better care through early and uniform diagnosis. Trial registration number No

The safety and robustness of large language models (LLMs) based applications remain critical challenges in artificial intelligence. Among the key threats to these applications are prompt hacking attacks, which can significantly undermine the security and reliability of LLM-based systems. In this work, we offer a comprehensive and systematic overview of three distinct types of prompt hacking: jailbreaking, leaking, and injection, addressing the nuances that differentiate them despite their overlapping characteristics. To enhance the evaluation of LLM-based applications, we propose a novel framework that categorizes LLM responses into five distinct classes, moving beyond the traditional binary classification. This approach provides more granular insights into the AI's behavior, improving diagnostic precision and enabling more targeted enhancements to the system's safety and robustness.

Maternal reports of decreased fetal movement (DFM) are a common reason to present to maternity care and are associated with stillbirth and other adverse outcomes. Promoting awareness of fetal movements and prompt assessment of DFM has been recommended to reduce stillbirths. However, evidence to guide clinical management of such presentations is limited. Educational approaches to increasing awareness of fetal movements in pregnant women and maternity care providers with the aim of reducing stillbirths have recently been evaluated in a several large clinical trials internationally. The International Stillbirth Alliance Virtual Conference in Sydney 2021 provided an opportunity for international experts in fetal movements to share reports on the findings of fetal movement awareness trials, consider evidence for biological mechanisms linking DFM and fetal death, appraise approaches to clinical assessment of DFM, and highlight research priorities in this area. Following this workshop summaries of the sessions prepared by the authors provide an overview of understandings of fetal movements in maternity care at the current time and highlights future directions in fetal movement research.

Routine management of patients classified with a pregnancy of unknown location (PUL) involves serial monitoring of hCG levels to triage women as being at high or low risk of complications. This study aimed to explore the feasibility and accuracy of capillary hCG point-of-care testing (POCT). In this prospective observational study women were invited to participate following a classification of PUL at initial ultrasound. At the time of routine blood collection for standard laboratory serum hCG testing (SLT), patients were offered venous and capillary hCG POCT using the Abbott i-STAT 1 device. Capillary samples were derived using a lancet finger prick and direct application of the bleb to the i-STAT 1 cartridge. Data was collected between April and December 2022 at a single London early pregnancy unit. 195 capillary POCT samples were collected and compared with SLT and venous POCT results. hCG readings > 2000 IU/L were excluded from analyses as the i-STAT 1 has an upper detection limit of 2000 IU/L. Venous POCT results did not differ significantly from SLT (p = .224). Capillary POCT values were significantly lower than venous POCT and SLT (p < .001) but showed a strong correlation with both SLT (r(132) = .97, p < .001) and venous POCT (r(124) = .99, p < 0.001). Capillary POCT showed even higher correlations with SLT when hCG levels were < 1000 IU/L (r(115) = .99, p < .001). The cartridge failure rate for the i-STAT 1 was 26% with capillary testing. Capillary POCT cannot be used interchangeably with standard laboratory derived results but has a high correlation with both venous POCT and laboratory testing. This suggests potential for future application as a simpler, faster, and less invasive alternative to venous blood sampling. Further work is required to reduce cartridge failure and calibrate existing PUL triaging models to allow safe use with capillary POCT.

Objectives: Women are increasingly opting for expectant care of ectopic pregnancy to avoid surgical intervention and potentially safeguard future fertility.Robust prognostic markers to predict which patients are likely to have successful conservative management are few.The objective of this study was to investigate the biochemical and sonographic characteristics of patients with ectopic pregnancy and determine which were predictive of successful expectant management.Methods: This retrospective cohort study examined demographic details, routine biochemistry and sonographic parameters for all cases of ectopic pregnancy across a 24-month period at a single London hospital trust.Results: The mean patient age was 32.3 (standard deviation = 5.7).In total, there were 219 cases of tubal ectopic pregnancy diagnosed on ultrasound.Of these, 95 had initial expectant management (43.4%) and were included in the study.The failure rate was 41.1% for expectant care (39 patients), the most common reason being a static or rising hCG.hCG at point of decision was significantly higher in those with failed versus successful expectant care (p < 0.01), as was the 48-hour hCG ratio (p = 0.013).Progesterone level did not significantly differ in those with successful versus failed expectant care, though in the overall cohort progesterone levels were significantly higher in patients requiring surgery at any point compared to those with resolution following expectant or medical care.Neither endometrial thickness nor presence of any free fluid were prognostic for failure of expectant management.Patients presenting initially as pregnancy of unknown location were no more likely to have successful management than those diagnosed with ectopic at the first visit.Conclusions: HCG at point of decision and 48-hour hCG ratio have the greatest utility in predicting success of expectant management for ectopic pregnancy.No other significant differences in clinical, biochemical or ultrasound findings were identified.

A randomised, parallel-group, double-blind, placebo-controlled multicentre study with health economic and nested qualitative studies to determine if mifepristone (Mifegyne®, Exelgyn, Paris, France) plus misoprostol is superior to misoprostol alone for the resolution of missed miscarriage. Women diagnosed with missed miscarriage in the first 14 weeks of pregnancy were randomly assigned (1 : 1 ratio) to receive 200 mg of oral mifepristone or matched placebo, followed by 800 μg of misoprostol 2 days later. A web-based randomisation system allocated the women to the two groups, with minimisation for age, body mass index, parity, gestational age, amount of bleeding and randomising centre. The primary outcome was failure to pass the gestational sac within 7 days after randomisation. The prespecified key secondary outcome was requirement for surgery to resolve the miscarriage. A within-trial cost-effectiveness study and a nested qualitative study were also conducted. Women who completed the trial protocol were purposively approached to take part in an interview to explore their satisfaction with and the acceptability of medical management of missed miscarriage. A total of 711 women, from 28 hospitals in the UK, were randomised to receive either mifepristone plus misoprostol (357 women) or placebo plus misoprostol (354 women). The follow-up rate for the primary outcome was 98% (696 out of 711 women). The risk of failure to pass the gestational sac within 7 days was 17% (59 out of 348 women) in the mifepristone plus misoprostol group, compared with 24% (82 out of 348 women) in the placebo plus misoprostol group (risk ratio 0.73, 95% confidence interval 0.54 to 0.98; p = 0.04). Surgical intervention to resolve the miscarriage was needed in 17% (62 out of 355 women) in the mifepristone plus misoprostol group, compared with 25% (87 out of 353 women) in the placebo plus misoprostol group (risk ratio 0.70, 95% confidence interval 0.52 to 0.94; p = 0.02). There was no evidence of a difference in the incidence of adverse events between the two groups. A total of 42 women, 19 in the mifepristone plus misoprostol group and 23 in the placebo plus misoprostol group, took part in an interview. Women appeared to have a preference for active management of their miscarriage. Overall, when women experienced care that supported their psychological well-being throughout the care pathway, and information was delivered in a skilled and sensitive manner such that women felt informed and in control, they were more likely to express satisfaction with medical management. The use of mifepristone and misoprostol showed an absolute effect difference of 6.6% (95% confidence interval 0.7% to 12.5%). The average cost per woman was lower in the mifepristone plus misoprostol group, with a cost saving of £182 (95% confidence interval £26 to £338). Therefore, the use of mifepristone and misoprostol for the medical management of a missed miscarriage dominated the use of misoprostol alone. The results from this trial are not generalisable to women diagnosed with incomplete miscarriage and the study does not allow for a comparison with expectant or surgical management of miscarriage. Future work should use existing data to assess and rank the relative clinical effectiveness and safety profiles for all methods of management of miscarriage. Our trial showed that pre-treatment with mifepristone followed by misoprostol resulted in a higher rate of resolution of missed miscarriage than misoprostol treatment alone. Women were largely satisfied with medical management of missed miscarriage and would choose it again. The mifepristone and misoprostol intervention was shown to be cost-effective in comparison to misoprostol alone. Current Controlled Trials ISRCTN17405024. This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 25, No. 68. See the NIHR Journals Library website for further project information.

Kisspeptin (KP), expressed in placental syncitiotrophoblasts, regulates physiological placentation, which is critical for pregnancy success. Circulating KP levels are raised in healthy pregnancy and reduced in miscarriage. We aimed to investigate the utility of measuring circulating KP in the assessment of pregnancy complications. This was a prospective cohort study conducted in an EPU in London from March 2013-2016. Women with intrauterine pregnancies (IPUV and viable pregnancies) were recruited and had ultrasound scan assessment and blood sampling every two weeks in the first trimester. Pregnancies were followed to final outcome and data collected using hospital records. Within this cohort, pregnancies examined included first trimester miscarriage (n = 95), pre-eclampsia (PET; n = 24), pregnancy induced hypertension (PIH; n = 14), gestational diabetes (GDM; n = 41), preterm birth (PTB; n = 14) and intrauterine growth restriction (IUGR; n = 24) (total of 569 samples). Asymptomatic healthy pregnancies were selected as controls (n = 265; samples n = 960). KP and beta human chorionic gonadotrophin (βhCG) levels were measured. Gestation-adjusted kisspeptin and βhCG levels were lower, by 66% and 57%, respectively, in women with miscarriage compared to healthy pregnant controls (p < 0.0001). To assess the performance of KP and βhCG to predict miscarriage, samples taken prior to miscarriage diagnosis were selected and compared with gestationally matched controls. A combination of both KP and βhCG performed best (AUC 0.881; 95% CI 0.834-0.927) when compared with either KP (AUC 0.845; 95% CI 0.803-0.882) or βhCG alone (AUC 0.806; 95% CI 0.749-0.863). Gestation-adjusted KP levels were lower in women with GDM (P = 0.002), and IUGR (P < 0.0001), and higher in women with PTB (P = 0.004). KP increased with gestation greater in PET (P = 0.008) and PIH (P < 0.0001) than in healthy controls. Plasma kisspeptin is a promising biomarker for prediction of first trimester miscarriage, and provides additional diagnostic capability over that provided by βhCG.

Risk prediction models utilising hCG and progesterone are effective tools to manage PUL pending a final diagnosis by TVUS. The M6 prediction model uses progesterone, hCG and the hCG ratio (hCG 48 hours/hCG 0 hours). Although often measured in clinical practice, there are limited data regarding the use of hCG measurements beyond 48 hours. We assessed the effectiveness of PUL risk stratification at 96 hours using the M6 model and the time taken to confirm a diagnosis of ectopic pregnancy using TVUS. This was a secondary analysis of 3977 prospective PUL cases at 8 UK hospitals between January 2015 and August 2018. PUL outcomes were defined as at low risk of clinical complications: intrauterine pregnancy [viable (VIUP) and non-viable (NVIUP)] or failed PUL (FPUL); or at high risk: persistent PUL (PPUL) and ectopic pregnancy (EP). Following 0 and 48-hour blood testing the M6 model protocol advises repeating the serum hCG with TVUS at 96 hours (i.e. a further 48 hours) if the PUL is classified as high risk. 466 patients predicted as high risk at 48 hours then had a 96-hour hCG measurement. This data was then analysed using the model. 466 PUL were classified as high risk at 48 hours: 118/466 EP (25.4%), 71/466 PPUL (15.2%), 54/466 VIUP (11.6%), 140/466 FPUL (30.0%) and 83/466 NVIUP (17.8%). At 96 hours, 372/466 PUL (79.8%) were classified as high risk and 94/466 (20.2%) low risk. 96-hour sensitivity of M6 for EP was 92.4% (109/118). All EP were visualised on follow-up TVUS and 84% within 8 days of presentation. Of 9 EP (7.6%) classified as low risk at 96 hours, 3 underwent operative intervention following diagnosis by TVUS at 96 hours. There were no ruptured EP. At 96 hours the M6 model has high sensitivity for EP whilst a further 20% PUL can be classified as low-risk, reducing unnecessary follow-up. After this time point TVUS visualised all EP in the cohort whilst serum hCG had limited value.