COMPARATIVE ANALYSIS OF MACHINE LEARNING ALGORITHMS FOR PROSTATE CANCER DETECTION

Prostate cancer currently continues to be a common disease thus need for early detection and efficient management of patients with this disease. Artificial intelligence (AI) or more specifically, Machine learning (ML) is capable to provide higher level of analytics solution on the large medical data or medical big data for higher diagnosis and prognostic results. Performance comparison of Support Vector Machines (SVM), Random Forests, k-Nearest Neighbors (k-NN), and Neural Network using comparatively large datasets ranging from Kaggle Prostate Cancer Dataset to SEER Database will be assessed in this study.The models were evaluated employing commonly recognized accuracy, sensitivity, specificity, as well as AUC parameters. The study found out that Random Forests had recorded the highest accuracy of 91.1% and AUC of 0.93 on the Kaggle dataset with good result on the SEER database. SVM came immediately behind with an accuracy of AUC = 0.92, which is considered accurate in clinical uses. These findings confirm the possibility of improving diagnostics of PC and creating a new vision of the perspective of using ensemble methods in modern medicine based on the development of ML models.