Development of a Predictive Model for Prostate Cancer Using a Machine Learning Based Classification Algorithm

Abstract

Prostate cancer remains one of the most prevalent malignancies among men worldwide, with early detection being crucial for effective treatment and improved survival outcomes. Traditional diagnostic procedures, such as prostate-specific antigen (PSA) testing, digital rectal examination (DRE), and biopsy, often suffer from limitations including subjectivity, low specificity, and inconsistent accuracy. This study presents the development of a predictive model for prostate cancer using a machine learning-based classification algorithm, specifically the Support Vector Machine (SVM). The dataset utilised was obtained from a publicly available prostate cancer repository, containing relevant biomedical and demographic features. Preprocessing procedures, including normalisation and data transformation, were applied to enhance model quality and ensure robustness. Experimental results revealed that the SVM model achieved a high predictive accuracy of 84.8% under crossvalidation and 87% on full dataset evaluation, with a corresponding error rate of less than 0.17. These results demonstrate the model’s ability to accurately distinguish between malignant and non-malignant cases, validating its suitability for clinical decision support. The model’s performance further confirms the potential of SVM as an effective classification technique for medical diagnostics, especially where datasets exhibit complex, nonlinear feature interactions. This study emphasises the significance of machine learning in enhancing diagnostic precision and reliability within the medical domain. The outcomes provide valuable insights into integrating artificial intelligence into healthcare systems for early cancer detection, reducing diagnostic delays, and supporting medical professionals in clinical decision-making.