NurtureNest: an IoT-wearable predictive analytics framework for real-time maternal risk assessment

Area of Science:

• Maternal Health
• Machine Learning in Healthcare
• Internet of Things (IoT)

Background:

• Global increase in pregnancy-related complications necessitates improved risk prediction.
• Current methods may lack continuous monitoring and timely intervention capabilities.

Purpose of the Study:

• To develop and evaluate an automated pregnancy risk assessment framework using IoT and machine learning.
• To enable continuous remote monitoring and early detection of high-risk pregnancies.

Main Methods:

• Integration of wearable sensor data (smartwatches) and clinical parameters via a mobile app.
• Training and evaluation of machine learning models (including ensemble techniques) for multi-class risk classification (low, medium, high).
• Validation using ROC analysis and feature importance.

Main Results:

• LightGBM model achieved 92.86% test accuracy and 95.04% cross-validation accuracy.
• The framework demonstrated effectiveness in classifying pregnancy risk levels.
• Feature importance analysis provided insights into key risk indicators.

Conclusions:

• The NurtureNest framework offers a promising approach for automated, continuous pregnancy risk assessment.
• Early risk detection supports timely clinical decision-making and improved maternal outcomes.
• IoT and machine learning integration can significantly advance remote maternal healthcare.