(1) Background: Preeclampsia (PE) prediction in the first trimester of pregnancy is a challenge for the clinicians. The aim of this study was to evaluate and compare the predictive performances of machine-learning based models for the prediction of preeclampsia, and its subtypes; (2) Methods: This prospective case-control study evaluated pregnancies that occurred in women who attended a tertiary maternity hospital in Romania between November 2019 and September 2022. The patients’ clinical and paraclinical characteristics were evaluated in the first trimester, and were included in 4 machine learning based models: decision tree (DT), naïve Bayes (NB), support vector machine (SVM), and random forest (RF), and their predictive performance was assessed; (3) Results: early-onset PE was best predicted by DT (accuracy: 94.1%), and SVM (accuracy: 91.2%) models, while NB (accuracy: 98.6%), and RF (accuracy: 92.8%) models had the highest performance when used to predict all types of PE. The predictive performance of these models was modest for moderate and severe types of PE, with accuracies ranging from 70.6% and 82.4%; (4) The machine learning-based models could be useful tools for PE prediction in the first trimester of pregnancy.

(1) Background: Neonatal early-onset sepsis (EOS) is associated with important mortality and morbidity. The aims of this study were to evaluate the association between serum biomarkers with early onset neonatal sepsis in a cohort of patients with prolonged rupture of membranes (PROM), and to calculate their diagnostic accuracy. (2) Methods: A retrospective cohort study was conducted on 1355 newborns with PROM admitted between January 2017 and March 2020, who were divided in two groups: group A, with PROM ≥ 18 hours and group B, with PROM < 18 hours. Both groups were further split into subgroups: proven sepsis, probable sepsis and no sepsis. Descriptive statistics, ANOVA analysis, and a Random Effects Generalized Least Squares (GLS) regression were used to evaluate the data. (3) Results: The statistically significant predictors of neonatal sepsis were the high white blood cells count from the first (p = 0.005) and third day (p = 0.028), and high CRP values from the first day (p = 0.004). Procalcitonin (area under the curve - AUC = 0.78) and CRP (AUC = 0.76) measured on the first day had the best predictive performance for early onset neonatal sepsis. (4) Conclusions: Further studies are needed to explore other bi-omarkers of sepsis in both pregnant women and newborns to better predict EOS.

1) Background: The screening of preeclampsia (PE) and intrauterine growth restriction (IUGR) represents a constant challenge for obstetricians. The aim of this study was to determine and compare the predictive performance of 4 machine learning-based algorithms for the prediction of PE, IUGR, and their association in a cohort of singleton pregnancies; (2) Methods This prospective study was conducted at a tertiary maternity hospital in Romania, and included 210 pregnancies that underwent first trimester screening. We included clinical and paraclinical data into 4 machine learning-based algorithms decision tree (DT), naïve Bayes (NB), support vector machine (SVM), and random forest (RF), and calculated their predictive performance; (3) Results: RF performed the best when used to predict PE, IUGR, and its subtypes, as well as the association between PE and IUGR. The overall predictive performance of DT for all these disorders was inferior to RF, NB, and SVM. Both SVM and NB had similar accuracy for the prediction of PE, while NB performed better than SVM for the prediction of IUGR; (4) Conclusions: Machine-learning-based algorithms could be useful for the prediction of ischemic placental disease and need to be validated on large cohorts of patients.

(1) Background: Urinary tract infections (UTIs) are an important cause of perinatal and maternal morbidity and mortality. The aim of this study was to describe and compare the main pregnancy outcomes among pregnant patients with complicated and uncomplicated UTIs; (2) Methods: This retrospective study included 183 pregnant patients who were evaluated for uncomplicated UTIs and urosepsis in the Urology Department of ‘C.I. Parhon’ University Hospital, and who were followed up at a tertiary maternity hospital- ‘Cuza-voda’ from Romania between January 2014 and October 2023. The control group (183 patients) was randomly selected from the patient’s cohort who gave birth in the same time frame at the maternity hospital without urinary pathology. Clinical and paraclinical data was examined. Descriptive statistics and a conditional logistic regression model were used to analyze our data. (3) Results: Our results indicated that patients with urosepsis had increased risk of premature rupture of membranes (aOR: 5.59, 95%CI: 2.02- 15.40, p&lt; 0.001) and preterm birth (aOR: 2.47, 95%CI: 1.15- 5.33, p= 0.02). We could not demonstrate a statistically significant association between intrauterine growth restriction and preeclampsia with the studied urological pathologies. (4) Conclusions: Careful UTIs screening during pregnancy is needed for preventing maternal-fetal complications.

(1) Background: The prediction of cervical lesions evolution is a challenge for clinicians. The prospective study aimed to determine and compare the predictive accuracy of cytology, HPV genotyping, and p16/Ki67 dual staining alone or in combination with personal risk factors for the prediction of progression, regression or persistence of cervical lesions in human papillomavirus (HPV)-infected patients; (2) Methods: This prospective study included HPV-positive patients with or without cervical lesions that underwent follow-up. Descriptive statistics and sensitivity analysis for individual or combined index tests were used for data analysis; (3) Results: The highest predictive performance for the progression of cervical lesions was achieved by a model that comprised a Pap smear suggestive of high-grade squamous intraepithelial lesion (HSIL), the presence of 16/18 HPV strains, a positive p16/Ki67 dual staining result along with the presence of at least 3 clinical risk factors, which had a sensitivity (Se) of 74.42%, a specificity of 97.92%, an area under the receiver operating curve (AUC) of 0.961, and an accuracy of 90.65%. The prediction of cervical lesions regression or persistence was modest; (4) Conclusions: Multiple testing or new biomarkers should be used to improve HPV-positive patient surveillance, especially for cervical lesion regression or persistence prediction.