Machine learning models for predicting preeclampsia: a systematic review.

BACKGROUND: This systematic review provides an overview of machine learning (ML) approaches for predicting preeclampsia. METHOD: This review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyzes (PRISMA) guidelines. We searched the Cochrane Central Register, PubMed, EMBASE, ProQuest, Scopus, and Google Scholar up to February 2023. Search terms were limited to "preeclampsia" AND "artificial intelligence" OR "machine learning" OR "deep learning." All studies that used ML-based analysis for predicting preeclampsia in pregnant women were considered. Non-English articles and those that are unrelated to the topic were excluded. The PROBAST was used to assess the risk of bias and applicability of each included study. RESULTS: The search strategy yielded 128 citations; after duplicates were removed and title and abstract screening was completed, 18 full-text articles were evaluated for eligibility. Four studies were included in this review. Two studies were at low risk of bias, and two had low to moderate risk. All of the study designs included were retrospective cohort studies. Nine distinct models were chosen as ML models from the four studies. Maternal characteristics, medical history, medication intake, obstetrical history, and laboratory and ultrasound findings obtained during prenatal care visits were candidate predictors to train the ML model. Elastic net, stochastic gradient boosting, extreme gradient boosting, and Random forest were among the best models to predict preeclampsia. All four studies used metrics such as the area under the curve, true positive rate, negative positive rate, accuracy, precision, recall, and F1 score. The AUC of ML models varied from 0.860 to 0.973 in four studies. CONCLUSION: The results of studies yielded high prediction performance of ML models for preeclampsia risk from routine early pregnancy information.

Machine learning-based approach for predicting low birth weight.

BACKGROUND: Low birth weight (LBW) has been linked to infant mortality. Predicting LBW is a valuable preventative tool and predictor of newborn health risks. The current study employed a machine learning model to predict LBW. METHODS: This study implemented predictive LBW models based on the data obtained from the "Iranian Maternal and Neonatal Network (IMaN Net)" from January 2020 to January 2022. Women with singleton pregnancies above the gestational age of 24 weeks were included. Exclusion criteria included multiple pregnancies and fetal anomalies. A predictive model was built using eight statistical learning models (logistic regression, decision tree classification, random forest classification, deep learning feedforward, extreme gradient boost model, light gradient boost model, support vector machine, and permutation feature classification with k-nearest neighbors). Expert opinion and prior observational cohorts were used to select candidate LBW predictors for all models. The area under the receiver operating characteristic curve (AUROC), accuracy, precision, recall, and F1 score were measured to evaluate their diagnostic performance. RESULTS: We found 1280 women with a recorded LBW out of 8853 deliveries, for a frequency of 14.5%. Deep learning (AUROC: 0.86), random forest classification (AUROC: 0.79), and extreme gradient boost classification (AUROC: 0.79) all have higher AUROC and perform better than others. When the other performance parameters of the models mentioned above with higher AUROC were compared, the extreme gradient boost model was the best model to predict LBW with an accuracy of 0.79, precision of 0.87, recall of 0.69, and F1 score of 0.77. According to the feature importance rank, gestational age and prior history of LBW were the top critical predictors. CONCLUSIONS: Although this study found that the extreme gradient boost model performed well in predicting LBW, more research is needed to make a better conclusion on the performance of ML models in predicting LBW.

Maternal and Neonatal Outcomes of Sub-clinical Hypothyroidism Treated With Levothyroxine in Pregnancy: A Retrospective Cohort Study.

INTRODUCTION: The effect of sub-clinical hypothyroidism (SCH) in pregnancy has been controversial. Furthermore, the impact of levothyroxine replacement on improving outcomes in pregnant women with SCH is unknown. This study aimed to assess the maternal and neonatal outcomes of pregnant women with SCH who were treated with levothyroxine replacement. METHODS: This retrospective chart review was conducted at a tertiary hospital in Iran between 2020 and 2022. All pregnant women who had given birth during the study period were recruited. Those who did not have thyroid function test results within 10-12 weeks, as well as those with SCH who did not have levothyroxine replacement, were excluded. The subjects were divided into two groups based on the 2017 American Thyroid Association (ATA) criteria: non-SCH (TSH values 0.27-2.5 mIU/L) and SCH (TSH values more than 4.0 mIU/L). The demographic, obstetric, maternal, and neonatal outcomes of both groups were compared. The Chi-square test was used to compare the categorical variables. Binary logistic regression was used to assess differences in categorical variables. RESULTS: With a frequency of 10.5%, 935 women out of 8888 were diagnosed with SCH. In terms of age, educational level, living residency, medical insurance, access to prenatal care, and smoking status, there were no differences between the two groups. In terms of gestational age, parity, onset of labor, history of infertility, hypertension, cardiovascular disease, anemia, and overt diabetes, there were no differences between the two groups; however, gestational diabetes was more common in those with SCH. Compared with the non-SCH group, the prevalence and risks of gestational diabetes [19.8 vs. 14.2, odds ratio (OR) = 1.14, 95% confidence interval (CI) = 1.72-3.95] were significantly higher in the SCH group after controlling for confounding factors. There were no differences in neonatal outcomes between the two groups. CONCLUSIONS: Except for gestational diabetes, we found no significant adverse events in terms of maternal and neonatal outcomes among women with SCH who were treated with levothyroxine.

Manifestation of Spontaneous Rupture of the Urinary Bladder in Pregnancy: A Systematic Review of the Literature.

Spontaneous rupture of the urinary bladder (SRUB) during pregnancy is a potentially fatal event that necessitates immediate surgery. The aim of this systematic review is to determine the symptoms, causes, associated factors, and prognosis of SRUB in pregnancy. We searched the literature from inception until December 2022 using the Cochrane Central Register, PubMed, EMBASE, ProQuest, Scopus, and Google Scholar. Articles not in English and those unrelated to the topic were excluded. The JBI Critical Appraisal Checklist for case reports was employed for the risk of bias assessment. The search strategy yielded 312 citations; 29 full-text articles were evaluated for eligibility after screening. Five case reports were included in this review. The age range of the cases was 27-39 years. Four cases were in their second trimester of pregnancy, and one was in her third. Two cases had undergone previous cesarean sections, and one had an appendectomy and salpingectomy due to an ectopic pregnancy. One case reported a history of alcohol and drug abuse. The most common symptoms were abdominal pain, abdominal distension, oliguria, voiding difficulty, hematuria, fever, and vomiting. The diagnosis of SRUB was primarily made via emergency laparotomy due to the patients' critical conditions. Beyond its diagnostic role, laparotomy was also the treatment method in all cases. Tear repair, antibiotic therapy, and urinary catheterization were all integral parts of the treatment. Four cases resulted in an uneventful pregnancy and a healthy, full-term baby. In one case, a hysterectomy was performed due to a combined uterus and bladder rupture. SRUB often presents with non-specific symptoms, leading to a delayed diagnosis. A high index of suspicion is essential when a pregnant woman exhibits urinary symptoms and severe abdominal pain, especially in those with a history of previous surgery. Early detection and treatment of SRUB are critical for an uneventful recovery.

Machine Learning-Based Approach to Predict Intrauterine Growth Restriction.

INTRODUCTION: Creating a prediction model incorporating multiple risk factors for intrauterine growth restriction is vital. The current study employed a machine learning model to predict intrauterine growth restriction. METHODS: This cross-sectional study was carried out in a tertiary hospital in Bandar Abbas, Iran, from January 2020 to January 2022. Women with singleton pregnancies above the gestational age of 24 weeks who gave birth during the study period were included. Exclusion criteria included multiple pregnancies and fetal anomalies. Four statistical learning algorithms were used to build a predictive model: (1) Decision Tree Classification, (2) Random Forest Classification, (3) Deep Learning, and (4) the Gradient Boost Algorithm. The candidate predictors of intrauterine growth restriction for all models were chosen based on expert opinion and prior observational cohorts. To investigate the performance of each algorithm, some parameters, including the area under the receiver operating characteristic curve (AUROC), accuracy, precision, and sensitivity, were assessed. RESULTS: Of 8683 women who gave birth during the study period, 712 were recorded as having intrauterine growth restriction, with a frequency of 8.19%. Comparing the performance parameters of different machine learning algorithms showed that among all four machine learning models, Deep Learning had the greatest performance to predict intrauterine growth restriction with an AUROC of 0.91 (95% confidence interval, 0.85-0.97). The importance of the variables revealed that drug addiction, previous history of intrauterine growth restriction, chronic hypertension, preeclampsia, maternal anemia, and COVID-19 were weighted factors in predicting intrauterine growth restriction. CONCLUSIONS: A machine learning model can be used to predict intrauterine growth restriction. The Deep Learning model is an accurate algorithm for predicting intrauterine growth restriction.

Maternal and Neonatal Outcomes of Iron Deficiency Anemia: A Retrospective Cohort Study.

INTRODUCTION: Understanding the outcomes of anemia in pregnancy is critical. Since no study has been conducted regarding the maternal and neonatal outcomes of iron-deficiency anemia in Hormozgan province of Iran, this study aims to assess the maternal and neonatal outcomes of iron-deficiency anemia in women who gave birth in Hormozgan province from January 2020 to January 2022. METHODS:  We retrospectively assessed all singleton pregnant women who gave birth at a tertiary hospital in Bandar Abbas, Hormozgan province, Iran, for two years. We divided all women into iron-deficiency anemic and non-iron-deficiency anemic women. Iron-deficiency anemia was defined as hemoglobin less than 10.5 mg/dl at the time of admission without any other hemoglobinopathy, such as sickle cell anemia or thalassemia. Using electronic patient records, data were extracted from the Iranian Maternal and Neonatal Network (IMaN Net), a valid national system. Since the information of birth under 24 weeks of gestation is not recorded in this system, we excluded all deliveries under 24 weeks of gestation. The outcome measures of the study were demographic factors (age, education, residency place, access to prenatal care, smoking), obstetrical factors (parity, labor induction, fetal presentation, mode of delivery), and maternal and neonatal outcomes (the incidence of preeclampsia, gestational diabetes, placenta abruption, postpartum hemorrhage, maternal need for blood transfusion, maternal need for intensive care unit, preterm birth, low birth weight, intrauterine growth retardation, birth asphyxia, stillbirth, and neonatal intensive care admission). Chi-square tests were used to compare differences between iron-deficiency anemic and non-iron-deficiency anemic women. Logistic regression models were used to assess the effect of iron-deficiency anemia on maternal and neonatal outcomes. The result was presented as odds ratio (OR) or adjusted odds ratio (aOR) after adjusting for covariates and a 95% confidence interval (CI).  Results: The incidence of iron-deficiency anemia was 2.97%. Education and residency were among the demographic factors that differed significantly between groups. Iron-deficiency anemia was more frequent in those with higher education and women who lived in rural areas. In terms of obstetrical factors, method of delivery was the only significantly different factor between groups. Iron-deficiency anemic mothers had substantially more instrumental deliveries than non-iron-deficiency anemic mothers (4.3% vs. 0.8%), while the incidence of cesarean section was lower. Based on logistic regression in terms of maternal and neonatal outcomes, iron-deficiency anemic women had a substantially higher risk of the need for maternal blood transfusion (aOR: 6.54, 95%CI: 4.72-8.15), postpartum hemorrhage (aOR: 1.54, 95%CI: 0.71-2.11), preterm birth (aOR: 0.98, 95%CI: 0.45-1.13), low birth weight (aOR: 1.04, 95%CI: 0.78-2.01), intrauterine growth retardation (aOR: 1.30, 95%CI: 0.99-2.10), and neonatal intensive care admission (aOR: 1.06, 95%CI: p.52-2.72), after adjusting for educational level, residency place, and method of delivery. CONCLUSIONS: Despite the higher incidence of postpartum hemorrhage and maternal blood transfusion, we found no increase in maternal intensive care unit admission risk. Regarding neonatal outcomes, iron-deficiency anemia was linked to preterm birth, low birth weight, intrauterine growth retardation, and neonatal intensive care admission.