Prediction of high-risk pregnancy based on machine learning algorithms.

Area of Science:

• Healthcare Informatics
• Applied Machine Learning
• Maternal Health Research

Background:

• Maternal health management requires efficient tools for identifying high-risk pregnancies.
• Existing methods may lack the speed and accuracy needed for timely intervention.
• The Maternal Health Risk Dataset (MHRD) from Bangladesh provides a basis for developing predictive models.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting high-risk pregnancies.
• To identify the most effective algorithm for maternal health risk assessment.
• To create a decision-support tool for healthcare professionals.

Main Methods:

• Utilized six machine learning algorithms: Multilayer Perceptron (MLP), Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM).
• Trained and tested models on the MHRD dataset comprising 1014 pregnant women.
• Evaluated model performance based on accuracy and prediction rates.

Main Results:

• The Multilayer Perceptron (MLP) algorithm demonstrated superior performance among the tested models.
• The developed MLP model achieved an overall accuracy of 82% in predicting pregnancy risks.
• The model showed a high accuracy of 91% specifically for identifying high-risk pregnancies.
• The model processed data at a rate of 500 data items per second, supported by an NVIDIA GPU RTX3050Ti.

Conclusions:

• Machine learning, particularly MLP, offers a powerful and efficient approach to predicting high-risk pregnancies.
• The developed model can serve as a valuable decision-support tool for improving maternal healthcare management.
• This study highlights the potential of AI in enhancing the early detection and management of pregnancy complications.