Forecasting Maternal Women's Health Risks using Random Forest Classifier

Abstract

Nowadays, maternal health during pregnancy is a major concern, especially in rural areas where risks are increased by a lack of medical experts and poor infrastructure. The lack of effective methods for predicting maternal health risks poses a significant challenge in maternal healthcare. Traditional approaches often fall short of accurately identifying and managing complications during pregnancy, leading to adverse outcomes. To overcome, this new technique can accurately assess medical data in order to forecast threats to maternal health. This research investigates machine learning algorithms' potential in predicting such risks using medical data. The study utilizes a carefully curated dataset called the Maternal Health Risk Dataset, encompassing parameters like age, blood pressure, glucose levels, body temperature, heart rate, and risk level. Through ensemble learning-based feature engineering, algorithms like Random Forest, XG Boost, Support Vector Classifier (SVC), Decision Tree, and Logistic Multiclass are assessed. Results show Random Forest achieving an impressive accuracy of 94.26%, indicating its potential in maternal health risk prediction, crucial for minimizing adverse outcomes and enabling timely interventions during pregnancy.

Topics & Concepts