Abstract :

Maternal And Fetal Health Monitoring Is A Critical Priority In Global Healthcare, Necessitating The Early Identification Of Complications Such As Fetal Hypoxia And Uterine Distress. Traditional Diagnostic Approaches Often Rely On The Manual Interpretation Of Cardiotocography (CTG) Signals, Which Is Highly Subjective, Errorprone, And Time-consuming. Furthermore, Standard Machine Learning Models Tend To Struggle With The Sequential Complexities Of Physiological Data And The Inherent Class Imbalance Present In Clinical Datasets, Where Normal Cases Vastly Outnumber Pathological Ones. To Address These Challenges, This Study Proposes DeepXPregNet, A Novel, Intelligent Pregnancy Health Risk Prediction Framework That Integrates Heterogeneous Deep Learning Topologies Enhanced By An Attention Mechanism And Explainable AI (XAI). The Proposed Approach Utilizes The Benchmark UCI Cardiotocography Dataset Containing 2,126 Fetal Heart Rate (FHR) And Uterine Contraction Features. The Methodology Begins With Robust Data Preprocessing, Utilizing The Synthetic Minority Oversampling Technique (SMOTE) To Synthetically Balance The Clinical Classes And Prevent Model Bias. The Core Architecture Combines 1D-Convolutional Neural Networks (CNN) For Automated Spatial And Morphological Feature Extraction (e.g., Detecting Decelerations And Variability) With Bi-Directional Long Short-Term Memory (Bi-LSTM) Networks To Capture Long-range Temporal Physiological Dependencies. Central To This Architecture Is A Multi-Head Attention Mechanism That Allows The System To Dynamically Focus On The Most Critical Monitoring Intervals. Unlike Traditional Black-box Models, This Framework Incorporates SHAP (Shapley Additive ExPlanations) To Interpret Predictions Visually, Highlighting Influential Features To Enhance Clinical Trust. Extensive Experiments Demonstrate That DeepXPregNet Significantly Outperforms Baseline Models, Achieving A Superior Accuracy Of 97.58% And A Near 100% Recall For Highrisk Fetuses. This Work Contributes To Predictive Healthcare Analytics By Offering A Robust, Scalable, And Interpretable Solution For Early Obstetric Risk Stratification.

Published:

21-7-2026

Issue:

Vol. 26 No. 7 (2026)

Page Nos:

1058-1066

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite