ACME: A Classification Model for Explaining the Risk of Preeclampsia Based on Bayesian Network Classifiers and a Non-Redundant Feature Selection Approach

Informatics, Journal Year: 2024, Volume and Issue: 11(2), P. 31 - 31

Published: May 17, 2024

While preeclampsia is the leading cause of maternal death in Guayas province (Ecuador), its causes have not yet been studied depth. The objective this research to build a Bayesian network classifier diagnose cases while facilitating understanding that generate disease. Data for years 2017 through 2023 were gathered retrospectively from medical histories patients treated at “IESS Los Ceibos” hospital Guayaquil, Ecuador. Naïve Bayes (NB), Chow–Liu Tree-Augmented (TANcl), and Semi (FSSJ) algorithms considered building explainable classification models. A proposed Non-Redundant Feature Selection approach (NoReFS) perform feature selection task. model trained with TANcl NoReFS was best them, an accuracy close 90%. According model, whose age above 35 years, severe vaginal infection, live rural area, use tobacco, family history diabetes, had personal hypertension are those high risk developing preeclampsia.

Language: Английский

---

Leveraging Machine Learning to Predict and Assess Disparities in Severe Maternal Morbidity in Maryland

Healthcare, Journal Year: 2025, Volume and Issue: 13(3), P. 284 - 284

Published: Jan. 31, 2025

Background: Severe maternal morbidity (SMM) is increasing in the United States. The main objective of this study to test use machine learning (ML) techniques develop models for predicting SMM during delivery hospitalizations Maryland. Secondarily, we examine disparities by key sociodemographic characteristics. Methods: We used linked State Inpatient Database (SID) and American Hospital Association (AHA) Annual Survey data from Maryland 2016–2019 (N = 261,226 hospitalizations). first estimated relative risks across factors (e.g., race, income, insurance, primary language). Then, fitted LASSO and, comparison, Logit with 75 18 features. selection features was based on clinical expert opinion, a literature review, statistical significance, computational resource constraints. Various model performance metrics, including area under receiver operating characteristic curve (AUC), accuracy, precision, recall values were computed compare predictive performance. Results: During 2016–2019, 76 per 10,000 deliveries (1976 261,226) patients who experienced an event. full list achieved AUC 0.71 validation dataset, which marginally decreased 0.69 reduced algorithm same demonstrated slightly superior 0.80. found significant among living low-income areas, public non-Hispanic Black or non-English speakers. Conclusion: Our results demonstrate feasibility utilizing ML administrative hospital discharge prediction. low score limitation all compared, signifying that algorithms struggle identifying cases. This identified substantial various factors. Addressing these requires multifaceted interventions include improving access quality care, enhancing cultural competence healthcare providers, implementing policies help mitigate social determinants health.

Language: Английский

---

Prediction of adverse pregnancy outcomes using machine learning techniques: evidence from analysis of electronic medical records data in Rwanda

BMC Medical Informatics and Decision Making, Journal Year: 2025, Volume and Issue: 25(1)

Published: Feb. 12, 2025

Despite substantial progress in maternal and neonatal health, Rwanda's mortality rates remain high, necessitating innovative approaches to meet health related Sustainable Development Goals (SDGs). By leveraging data collected from Electronic Medical Records, this study explores the application of machine learning models predict adverse pregnancy outcomes, thereby improving risk assessment enhancing care delivery. This utilized retrospective cohort electronic medical record (EMR) system 25 hospitals Rwanda 2020 2023. The independent variables included socioeconomic status, reproductive pregnancy-related factors. outcome variable was a binary composite feature that combined outcomes both mother newborn. Extensive cleaning performed, with missing values addressed through various strategies, including exclusion instances, imputation techniques using K-Nearest Neighbors Multiple Imputation by Chained Equations. Data imbalance managed synthetic minority oversampling technique. Six models—Logistic Regression, Decision Trees, Support Vector Machine, Gradient Boosting, Random Forest, Multilayer Perceptron—were trained 10-fold cross-validation evaluated on an unseen dataset with–70 − 30 training evaluation splits. 117,069 women across were analyzed, leading final 32,783 after removing entries significant values. Among these women, 5,424 (16.5%) experienced outcomes. Forest Boosting Classifiers demonstrated high accuracy precision. After hyperparameter tuning, model achieved 90.6% ROC-AUC score 0.85, underscoring its effectiveness predicting However, recall rate 46.5% suggests challenges detecting all cases. Key predictors identified gestational age, number pregnancies, antenatal visits, vital signs, delivery methods. recommends EMR quality, integrating into routine practice, conducting further research refine predictive address evolving In addition, design AI-based interventions for high-risk pregnancies. Not applicable.

Language: Английский

---

Maternal Health Risk Detection: Advancing Midwifery with Artificial Intelligence

Healthcare, Journal Year: 2025, Volume and Issue: 13(7), P. 833 - 833

Published: April 6, 2025

Background/Objectives: Maternal health risks remain one of the critical challenges in world, contributing much to maternal and infant morbidity mortality, especially most vulnerable populations. In modern era, with recent progress area artificial intelligence machine learning, promise has emerged regard achieving goal early risk detection its management. This research is set out relate high-risk, low-risk, mid-risk using learning algorithms based on physiological data. Materials Methods: The applied dataset contains 1014 instances (i.e., cases) seven attributes variables), namely, Age, SystolicBP, DiastolicBP, BS, BodyTemp, HeartRate, RiskLevel. preprocessed used was then trained tested six classifiers 10-fold cross-validation. Finally, performance metrics models erre compared like Accuracy, Precision, True Positive Rate. Results: best found for Random Forest, also reaching highest values Accuracy (88.03%), TP Rate (88%), Precision (88.10%), showing robustness handling classification. category challenging across all models, characterized by lowered Recall scores, hence underlining class imbalance as bottlenecks performance. Conclusions: Machine hold strong potential improving prediction. findings underline place advancing healthcare driving more data-driven personalized approaches.

Language: Английский

---

Ensemble machine learning framework for predicting maternal health risk during pregnancy

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Sept. 14, 2024

Language: Английский

---

Performance of machine‐learning approach for prediction of pre‐eclampsia in a middle‐income country

Ultrasound in Obstetrics and Gynecology, Journal Year: 2023, Volume and Issue: 63(3), P. 350 - 357

Published: Sept. 29, 2023

ABSTRACT Objective Pre‐eclampsia (PE) is a serious complication of pregnancy associated with maternal and fetal morbidity mortality. As current prediction models have limitations may not be applicable in resource‐limited settings, we aimed to develop machine‐learning (ML) algorithm that offers potential solution for developing accurate efficient first‐trimester PE. Methods We conducted prospective cohort study Mexico City, model preterm PE (pPE) using ML. Maternal characteristics locally derived multiples the median (MoM) values mean arterial pressure, uterine artery pulsatility index serum placental growth factor were used variable selection. The dataset was split into training, validation test sets. An elastic‐net method employed predictor selection, performance evaluated area under receiver‐operating‐characteristics curve (AUC) detection rates (DR) at 10% false‐positive (FPR). Results final analysis included 3050 pregnant women, whom 124 (4.07%) developed ML showed good performance, AUCs 0.897, 0.963 0.778 pPE, early‐onset (ePE) any type (all‐PE), respectively. DRs FPR 76.5%, 88.2% 50.1% ePE all‐PE, Conclusions Our demonstrated high accuracy predicting pPE MoM. provide an accessible tool early PE, facilitating timely intervention improved outcome. © 2023 Authors. Ultrasound Obstetrics & Gynecology published by John Wiley Sons Ltd on behalf International Society Gynecology.

Language: Английский

---

Validation of the first‐trimester machine learning model for predicting pre‐eclampsia in an Asian population

International Journal of Gynecology & Obstetrics, Journal Year: 2024, Volume and Issue: 167(1), P. 350 - 359

Published: April 26, 2024

To evaluate the performance of an artificial intelligence (AI) and machine learning (ML) model for first-trimester screening pre-eclampsia in a large Asian population.

Language: Английский

---

Novel Associations Between Mid-Pregnancy Cardiovascular Biomarkers and Preeclampsia: An Explorative Nested Case-Control Study

Reproductive Sciences, Journal Year: 2024, Volume and Issue: 31(5), P. 1391 - 1400

Published: Jan. 22, 2024

Abstract Prediction of women at high risk preeclampsia is important for prevention and increased surveillance the disease. Current prediction models need improvement, particularly with regard to late-onset preeclampsia. Preeclampsia shares pathophysiological entities cardiovascular disease; thus, biomarkers may contribute improving models. In this nested case-control study, we explored predictive importance mid-pregnancy subsequent We included healthy singleton pregnancies who had donated blood in (~ 18 weeks’ gestation). Cases were ( n = 296, 10% whom early-onset [< 34 weeks]). Controls 333). collected data on maternal, pregnancy, infant characteristics from medical records. used Olink II panel immunoassay measure 92 plasma samples. The Boruta algorithm was determine investigated first-trimester pregnancy development following confirmed associations (in descending order importance): placental growth factor (PlGF), matrix metalloproteinase (MMP-12), lectin-like oxidized LDL receptor 1, carcinoembryonic antigen-related cell adhesion molecule 8, serine protease 27, pro-interleukin-16, poly (ADP-ribose) polymerase 1. that associated BNP, MMP-12, alpha-L-iduronidase (IDUA), PlGF, low-affinity immunoglobulin gamma Fc region II-b, T surface glycoprotein. Our results suggest MMP-12 a promising novel biomarker. Moreover, BNP IDUA be value enhancing

Language: Английский

---

Prediction of childbearing tendency in women on the verge of marriage using machine learning techniques

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Sept. 6, 2024

The declining fertility rate and increasing marriage age among girls pose challenges for policymakers, leading to issues such as population decline, higher social economic costs, reduced labor productivity. Using machine learning (ML) techniques predict the desire have children can offer a promising solution address these challenges. Therefore, this study aimed childbearing tendency in women on verge of using ML techniques. Data from 252 participants (203 expressing "desire children" 49 indicating "reluctance children") Abadan, Khorramshahr cities (Khuzestan Province, Iran) was analyzed. Seven algorithms, including multilayer perceptron (MLP), support vector (SVM), logistic regression (LR), random forest (RF), J48 decision tree, Naive Bayes (NB), K-nearest neighbors (KNN), were employed. performance algorithms assessed metrics derived confusion matrix. RF algorithm showed superior performance, with highest sensitivity (99.5%), specificity (95.6%), receiver operating characteristic curve (90.1%) values. Meanwhile, MLP emerged top-performing algorithm, showcasing best overall accuracy (77.75%) precision (81.8%) compared other algorithms. Factors marriage, place residence, strength family center birth child most effective predictors woman's children. Conversely, number daughters, wife's ethnicity, spouse's ownership assets cars houses least important factors predicting desire. exhibit excellent predictive capabilities tendencies highlighting their remarkable effectiveness. This capacity accurate prognoses holds significant promise advancing research field.

Language: Английский

---

The association between first trimester blood pressure, blood pressure trajectory, mid-pregnancy blood pressure drop and maternal and fetal outcomes: A systematic review and meta-analysis

Pregnancy Hypertension, Journal Year: 2024, Volume and Issue: 38, P. 101164 - 101164

Published: Oct. 16, 2024

Hypertensive disorders of pregnancy occur in 5-10 % pregnancies and are associated with an increased risk adverse perinatal outcomes.

Language: Английский

---