A Comparison of Systematic, Targeted, and Combined Biopsy Using Machine Learning for Prediction of Prostate Cancer...

Area of Science:

• Oncology
• Medical Informatics
• Biostatistics

Background:

• Prostate cancer (PCa) diagnosis relies on various clinical factors and biopsy techniques.
• Accurate prognostic prediction models are crucial for timely and effective PCa management.
• Comparing machine learning (ML) model performance across different biopsy strategies is essential for optimizing detection.

Purpose of the Study:

• To develop novel machine learning (ML) models for predicting prostate cancer (PCa) risk.
• To evaluate and compare the efficacy of these ML models using systematic versus targeted biopsy detection techniques.

Main Methods:

• Analysis of 528 patients' data diagnosed with PCa from 2019-2023 in Riyadh, Saudi Arabia.
• Utilized four machine learning (ML) algorithms, including Random Forest (RF) and XGBoost (XGB), for PCa prediction and classification.
• Evaluated model performance based on factors like prostate-specific antigen (PSA), MRI findings, and biopsy type.

Main Results:

• Age, prostate volume, PSA, BMI, mpMRI score, and lesion characteristics were significantly associated with PCa.
• Random Forest (RF) and XGBoost (XGB) models demonstrated high accuracy in predicting PCa.
• Models achieved superior performance (AUC 0.94-0.97) for targeted and combined biopsies compared to systematic biopsy alone.

Conclusions:

• The Random Forest (RF) model exhibits excellent predictive capability for PCa risk, particularly with targeted and combined biopsy approaches.
• Machine learning (ML) models show promise as screening tools to potentially reduce missed PCa diagnoses.