GDVM: A Deep Learning Assisted Fetal Health Prediction System Using Gradient Distance Vector Model

The prediction of fetal health conditions is a critical aspect of prenatal care, and accurate detection of potential complications can greatly improve clinical outcomes. This paper proposes a novel Gradient Distance Vector Model (GDVM) integrated with a hybrid deep learning architecture for fetal health prediction. The model combines dimensionality reduction using a Hybrid AutoEncoder, temporal sequence modeling via Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU), and classification with Extreme Gradient Boosting (XGBoost). The proposed GDVM model was evaluated on publicly available fetal health datasets and achieved an impressive accuracy of 97.81%, outperforming nine existing models, including Random Forest, Support Vector Machine, and LSTM + GRU. In addition to superior accuracy, the GDVM model demonstrated excellent performance across other metrics, such as precision, recall, and F1-score. The model's ability to handle high-dimensional data and capture temporal patterns in fetal health metrics makes it a robust tool for predicting normal, suspected, and pathological fetal conditions. This approach offers significant potential for improving prenatal monitoring and early diagnosis of fetal complications.