Prostate Cancer Risk Prediction and Online Calculation Based on Machine Learning Algorithm.

Area of Science:

• Oncology
• Medical Informatics
• Machine Learning

Background:

• Prostate cancer (PCa) risk prediction is crucial for timely diagnosis and treatment.
• Evaluating artificial intelligence (AI) under healthcare data platforms offers new diagnostic avenues.

Purpose of the Study:

• To develop a PCa risk prediction model using common clinical indicators.
• To assess the efficacy of AI technologies within healthcare data platforms for PCa risk assessment.

Main Methods:

• Feature selection using Smoothly Clipped Absolute Deviation (SCAD).
• Development and comparison of Random Forest (RF), Support Vector Machine (SVM), Back Propagation Neural Network (BP), and Convolutional Neural Network (CNN) models.
• Utilized SMOTE for data enhancement in BP and CNN models.
• Performance evaluation via Area Under the Curve (AUC) of the Receiver Operating Characteristic curve.
• Developed an online risk prediction calculator using the Shiny platform.

Main Results:

• Inorganic phosphorus, triglycerides, and calcium were identified as significant indicators for PCa risk, alongside tissue volume and free prostate-specific antigen (PSA).
• The RF model demonstrated superior performance with an accuracy of 96.80% and an AUC of 0.975.
• BP and SVM models showed moderate performance, while CNN performed less effectively.
• An online PCa risk prediction tool was successfully developed based on the optimal RF model.

Conclusions:

• AI models, including traditional machine learning and deep learning, hold significant value for disease risk prediction on healthcare data platforms.
• The study proposes novel approaches for PCa risk prediction in patients suspected of having the disease, particularly those who have undergone core needle biopsy.
• The developed online calculator enhances the practical application of AI in medical diagnosis for PCa.