Maternal Health Risk Detection: Advancing Midwifery with Artificial Intelligence.

Area of Science:

• Maternal Health
• Artificial Intelligence
• Machine Learning

Background:

• Maternal health risks are a significant global challenge, contributing to morbidity and mortality.
• Artificial intelligence and machine learning offer promising solutions for early risk detection and management.
• Vulnerable populations are disproportionately affected by maternal health complications.

Purpose of the Study:

• To classify maternal health risk levels (high, mid, low) using machine learning algorithms.
• To analyze the effectiveness of different machine learning models on physiological data for risk prediction.
• To identify the best-performing algorithm for maternal health risk assessment.

Main Methods:

• Utilized a dataset of 1014 instances with seven physiological attributes (Age, SystolicBP, DiastolicBP, BS, BodyTemp, HeartRate).
• Trained and tested six classifiers with 10-fold cross-validation on preprocessed data.
• Evaluated model performance using Accuracy, Precision, and True Positive Rate.

Main Results:

• Random Forest classifier demonstrated the highest performance with 88.03% Accuracy, 88.10% Precision, and 88% True Positive Rate.
• The mid-risk category presented classification challenges, indicated by lower Recall and Precision scores.
• Class imbalance was identified as a key factor affecting model performance.

Conclusions:

• Machine learning algorithms show significant potential for enhancing maternal health risk prediction.
• Data-driven and personalized approaches in maternal healthcare can be advanced through machine learning.
• Further research is needed to address class imbalance for improved mid-risk category prediction.