PROSTATE CANCER DIAGNOSIS USING RESAMPLING TECHNOLOGY AND EXTRA TREES

Prostate cancer represents a significant health concern for middle-aged and elderly men in China, where both incidence rates are climbing and detection and treatment methodologies trail behind those of Western countries. This study aims to enhance the diagnostic and treatment processes for prostate cancer through the application of machine learning techniques. The research concentrates on enriching and equilibrating medical datasets using the Extra Trees prediction model. A comparative analysis confirms the effectiveness of this model in developing a robust diagnostic model for prostate cancer, thus supporting clinical decision-making. The study is divided into three main areas: first, machine learning techniques are applied to address missing values in prostate cancer datasets, with various approaches including Extra Trees and optimized ensemble algorithms assessed for effectiveness. Secondly, a strategy to balance datasets with mixed features is explored, particularly focusing on correcting imbalances in diagnostic outcomes, with random oversampling noted to enhance diagnostic precision. Finally, a diagnostic model based on the Extra Trees classifier is constructed and benchmarked against alternative algorithms and pertinent studies, establishing its suitability for prostate cancer diagnosis.