Improving Prostate Cancer Classification with Random Forest Techniques

Abstract

Prostate cancer is a leading cause of cancer-related mortality among men worldwide, necessitating accurate and efficient classification methods for improved diagnosis and treatment planning. This research explores the application of Random Forest algorithms to classify prostate cancer cases using a dataset comprising 100 samples with features such as radius, texture, perimeter, area, smoothness, compactness, symmetry, and fractal dimension. The study emphasizes the integration of preprocessing, feature selection, model training, and evaluation to enhance classification performance. The model achieved a classification accuracy of 75%, with a high recall of 88% for malignant cases, demonstrating its potential in identifying high-risk patients. However, the model exhibited challenges in predicting benign cases due to class imbalance, as reflected in the low precision (33%) for this minority class. Addressing these limitations, techniques such as data balancing, advanced hyperparameter tuning, and enhanced feature engineering are suggested. This study provides valuable insights into key predictors of prostate cancer and highlights the potential of Random Forest techniques as a robust tool for clinical decision-making. Future work should focus on integrating additional clinical and genomic data to further improve classification accuracy and interpretability.