IoT Based Fetal Heathcare Prediction using Machine Learning Approaches

Abstract

Introduction: A fast-growing topic called "Internet-of-Things(IoT)/Machine Learning(ML) for Fetus Medical Prediction" uses a variety of AI (Artificial Intelligence) and information analytics approaches to enhance care during pregnancy. This entails gathering and analyzing information from many sources to forecast future health concerns for the fetus, which can assist medical professionals in taking preventative or management actions.

Objectives: The objective is to examine the IoT based fetal healthcare prediction with the IoT sensor’s data for the early detection of diseases affecting the fetus and preventing it.

Methods: This study explored the various ML algorithms including DT (Decision Tree), RF (Random Forest), SVM (Support Vector Machine), XGBoost, Adaptive Boosting, R-SVM, Decision Stump Model, and K Nearest Neighbour (KNN).

Results: Demonstrates the effectiveness of approaches in predicting fetal healthcare outcomes, offering promising results using the evaluation metrics. Through experimentation, we have demonstrated that the combination of RF with Min-Max scaling and Relief feature extraction stands out as a superior approach in terms of predictive performance. This combination not only enhances accuracy but also improves precision, recall, and F1-Score, thus offering a comprehensive evaluation of fetal healthcare outcomes.

Conclusions: IoT based fetal healthcare prediction utilizing a diverse range of machine learning algorithms and pre-processing techniques has yielded insightful findings. By leveraging the strengths of Random Forest's ensemble learning, Min-Max scaling's normalization benefits, and Relief feature extraction's capability to select relevant features, our approach achieves a well-balanced prediction model that effectively captures intricate patterns within the fetal healthcare dataset.