9th International Conference on ICT for Sustainable Development

Authors - Rudrani Chavarkar, Shreya Kamath, Kumkum Saxena
Abstract - Uterine fibroids, common among women of childbearing age, cause major health challenges, including infertility and miscarriage. This study evaluates five machine learning algorithms, including Support Vector Machine, VGG16, Gaussian Mixture Model, Gaussian Mixture Model with Stacking (using Random Forest and Logistic Regression), and Random Forest for fibroid detection. Random Forest came forth as the best-performing model, achieving 99.0% accuracy with low inference time. The accuracy of SVM and VGG16 turns out to be 93.5% and 93.7% respectively but with certain limitations in precision and recall rates. Using the stacking ensembling technique in GMM-S we improved the performance resulting in an accuracy of 97.2%, which is closer to that of Random Forest while improving recall and precision rates. The findings of these machine learning approaches mark their capacity to improve the diagnostic accuracy of uterine fibroids.