<![CDATA[Maternal Mortality Rate (MMR) in Indonesia remains a significant health issue, with data indicating a mortality rate far exceeding the Sustainable Development Goals (SDGs) target. This study aimed to explore and compare the performance of K-Nearest Neighbors (KNN), Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM) algorithms in detecting maternal mortality risk. Using a medical dataset of pregnant women from Sumbersari Community Health Center, models were developed to classify three pregnancy risk categories: low risk (KRR), high risk (KRT), and very high risk (KRST). Model evaluation was conducted based on accuracy, precision, recall, and F1-score metrics. The results showed that the Random Forest algorithm achieved the highest performance with an accuracy of 76.7%, followed by Decision Tree and SVM with 70%, while KNN had the lowest accuracy at 50%. The main challenge encountered was data imbalance in the classification of very high-risk cases. This study suggests the use of data balancing methods such as SMOTE and additional data augmentation to enhance model performance. These findings can serve as a foundation for Puskesmas to implement machine learning-based early detection systems to reduce maternal mortality rates.]]>
Copyrights © 2025
