Machine Learning Techniques for Predicting Ovarian Cancer in its Early Stages Using Biomarkers

The deadliest gynecological cancer affecting women is ovarian cancer, currently incurable with no effective medication treatments. The key focus of this research is to assess insights for early diagnosis using statistical analysis and machine learning techniques on data from clinical trials obtained from 349 patients. Several techniques, including Random Forest, Decision Tree, Gaussian NB, AdaBoost, and Logistic regression, were applied to find the most reliable factor for ovarian cancer prediction. A clinically evaluated raw dataset of benign samples and malignant ovarian tumor patient data set is used to develop early-stage ovarian cancer predictions, and the effectiveness of ML models was examined utilizing metrics including F1-score, Accuracy, Precision, and Recall. The proposed study shows better outcomes, with the Random Forest classifier exhibiting the highest accuracy for validation at 99% based on the test data of ovarian cancer predictions. Even though early-stage ovarian cancer detection is generally unavailable, cancer diagnosis may be greatly aided by machine learning.