HIGH-PRECISION AI PIPELINE EXPLAINED IN PROSTATA CANCER

Authors

Keywords: prostate cancer, machine learning, feature engineering, PCA, SHAP, diagnostics

Abstract

This study presents a comprehensive feature engineering process for the early detection of prostate cancer using machine learning methodology. The dataset consisted of key clinical indicators — PSA level, patient age, prostate volume, Gleason score, and clinical stage — which were processed using ANOVA, Chi-square, PCA, RFE, LASSO, and SHAP techniques. The primary objective was to identify the most influential diagnostic features that improve model performance and ensure interpretability.

References

1. Wong M.C.S., Goggins W., Wang H.H. Global epidemiological trends of prostate cancer. // Cancer. 2022. Vol. 128(10). – P. 2104-2118.

2. Qu Y., Lu X., Yin P. Prostate cancer detection via explainable machine learning. // Computers in Biology and Medicine. 2023. Vol. 165. – P. 107-121.

3. Chen R., Liu Y., Qi J. PSA-based early diagnosis of prostate cancer using ML models. // BMC Urology. 2023. Vol. 23(18). – P. 1-10.

4. Алексеев Б.Я., Каприн А.Д. Рак предстательной железы: диагностика и лечение. – Москва: «Медицина», 2021.

5. Гладышев М.Ю., Лопатин В.Н. Применение машинного обучения в онкоурологии. // Онкоурология. 2022. Т. 18(3). – С. 41-48.

6. Трухин Д.А. Анализ факторов риска рака простаты на основе PSA и клинических данных. // Вестник Урологии. 2023. №2. – С. 55-62.