Improving prediction of maternal health risks using PCA features and TreeNet model

Area of Science:

• Public Health
• Biomedical Informatics
• Machine Learning in Healthcare

Background:

• Maternal healthcare is crucial for the well-being of mothers and fetuses.
• Pregnancy and postpartum periods present significant health risks.
• Timely detection of maternal health risks is vital for patient safety.

Purpose of the Study:

• To propose and evaluate an approach for predicting maternal health risks.
• To enhance the accuracy and efficiency of risk identification in maternal healthcare.

Main Methods:

• Utilized Principal Component Analysis (PCA) for feature extraction.
• Employed a stacked ensemble voting classifier combining machine learning and deep learning models.
• Compared the proposed model against six machine learning and one deep learning algorithm.

Main Results:

• The PCA-based approach achieved 98.25% accuracy, 99.17% precision, 99.16% recall, and 99.16% F1 score.
• The proposed model demonstrated superior performance compared to existing state-of-the-art methods.
• PCA-based features significantly improved the model's predictive capabilities.

Conclusions:

• The developed approach effectively predicts maternal health risks with high accuracy.
• PCA combined with stacked ensemble models offers a promising tool for improving maternal healthcare outcomes.
• This method can aid in the early identification and management of maternal health complications.