Home / Current Issue / Paper 1714356
Machine Learning-Based Prediction of Elevated Prostate-Specific Antigen Levels from Lifestyle and Demographic Data
Subject area: Science,Engineering and Technology · Area of research: Machine Learning in Prostate Screening
Abstract
Background: Machine learning offers promising approaches for medical prediction tasks. This study evaluates the comparative performance of three ML algorithms: Logistic Regression with L1 regularization, Support Vector Machine (SVM), and Random Forest in predicting elevated prostate-specific antigen (PSA) levels using lifestyle and demographic features. Objectives: To compare the predictive performance, generalization capability, and stability of multiple ML models for detecting elevated PSA levels via binary classification of PSA status. Methods: We implemented three ML algorithms with two feature selection approaches to address the events-per-variable (EPV) problem. Lifestyle and demographic data were collected from adult males in Etsako West LGA, Edo State, Nigeria. Models were trained on 70% of the data (n=69) and validated on 30% (n=30). Performance was assessed using accuracy, precision, recall, specificity, F1-score, and ROC-AUC. Five-fold stratified cross-validation evaluated model stability. Hyperparameter optimization was performed using GridSearchCV. Results: The Random Forest model achieved the most balanced performance with 73.3% accuracy, 70.6% precision, 80.0% recall, 63.6% specificity, 0.750 F1-score, and 0.764 ROC-AUC. SVM showed identical test set performance (73.3% accuracy, 0.764 ROC-AUC). Logistic Regression with L1 regularization and 11 features achieved the highest recall (100%) but at the cost of zero specificity, indicating overfitting. Cross-validation revealed model stability: Random Forest CV recall 0.833 ± 0.061, CV F1 0.813 ± 0.053. The F1-optimized Random Forest showed improved balance (70.0% accuracy, 66.7% recall, 50.0% specificity). All models demonstrated ROC-AUC between 0.714–0.764, indicating acceptable discrimination capability. Conclusions: Random Forest and SVM demonstrated the most balanced performance in terms of sensitivity and specificity for PSA prediction in a small-sample setting. The study highlights important methodological considerations, including the need for feature selection under EPV constraints, the role of regularization in mitigating overfitting, and the importance of cross-validation for evaluating models out of sample performance.The moderate performance (ROC-AUC ≈ 0.71–0.76) suggests that lifestyle-based ML models may be useful for preliminary screening but are not suitable for diagnostic applications. Future work should focus on larger datasets, external validation, and ensemble methods.
References
[1] Omankwu, O. C., Osodeke, E. C., & Kanu, C. (2023). Machine learning algorithms for predicting high-risk of prostate cancer using prostate-specific antigen (PSA), age and body mass index (BMI). NIPES Journal of Science and Technology Research, 5(1), 133–149. https://doi.org/10.5281/zenodo.7729236
[2] Mustafa Sungur, Aykut Aykaç, M. E. Aydin, Ö. Celik, & Coskun Kaya. (2024). Machine learning-based prediction of prostate biopsy necessity using PSA, MRI, and hematologic parameters. Journal of Clinical Medicine, 14(1), 183. https://doi.org/10.3390/jcm14010183
[3] Chia-Cheng Chang, J. K. Chiou, C.-J. Lin, K. Lu, J.-R. Li, L.-W. Chang, et al. (2024). Machine-learning algorithm-based risk prediction and screening-detected prostate cancer in a benign prostate hyperplasia cohort. Anticancer Research, 44(4), 1683–1693. https://doi.org/10.21873/anticanres.16967
[4] Mostafa A. Arafa, K. H. Farhat, S. F. Aly, F. K. Khan, A. Mokhtar, A. M. Althunayan, et al. (2025). Prediction of prostate biopsy outcomes at different cut-offs of prostate-specific antigen using machine learning: A multicenter study. Journal of the Egypt National Cancer Institute, 37(1), 8. https://doi.org/10.1186/s43046-025-00265-3
[5] Atilla Satır, Y. Üstündağ, M. R. Yeşil, & K. Huysal. (2025). Prediction of prostate cancer from routine laboratory markers with automated machine learning. Journal of Clinical Laboratory Analysis, 39(3), e25143. https://doi.org/10.1002/jcla.25143
[6] Emre Alataş, H. T. Kökkülünk, H. Tanyıldızı, & G. Alcın. (2025). Treatment prediction with machine learning in prostate cancer patients. Computer Methods in Biomechanics and Biomedical Engineering, 28(4), 572–580. https://doi.org/10.1080/10255842.2023.2298364
[7] Yunxun Liu, J. Wu, X. Ni, Q. Zheng, J. Wang, H. Shen, L. Wang, R. Yang, & X. Weng. (2025). Machine learning based on automated 3D radiomics features to classify prostate cancer in patients with PSA levels of 4–10 ng/mL. Translational Andrology and Urology, 14(4), 1025–1035. https://doi.org/10.21037/tau-2024-731
[8] Hernandez, K. A. (2024). Prostate cancer prediction using machine learning techniques. International Journal of Open Science and Innovation, 8(2), 87–94. https://doi.org/10.6977/IJoSI.202406_8(2).0007
[9] Imran, M., Brisbane, W. G., Su, L.-M., Joseph, J. P., & Shao, W. (2025). Prostate cancer screening with artificial intelligence-enhanced micro-ultrasound: A comparative study with traditional methods. arXiv Preprint. https://arxiv.org/abs/2505.21355
[10] Noorul Wahab, E. Alzaid, J. Lv, A. Shephard, S. E. A. Raza, & J. Ha. (2025). MultiSurv: A multimodal deep survival framework for prostate and bladder cancer. arXiv Preprint. https://arxiv.org/abs/2509.05037
How to cite this paper
@article{1714356,
author = {Onitcha Nyerhovwo Edafetanure},
title = {Machine Learning-Based Prediction of Elevated Prostate-Specific Antigen Levels from Lifestyle and Demographic Data},
journal = {Iconic Research And Engineering Journals},
year = {2025},
volume = {8},
number = {9},
pages = {1857-1866},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1714356.pdf},
abstract = {Background: Machine learning offers promising approaches for medical prediction tasks. This study evaluates the comparative performance of three ML algorithms: Logistic Regression with L1 regularization, Support Vector Machine (SVM), and Random Forest in predicting elevated prostate-specific antigen (PSA) levels using lifestyle and demographic features.
Objectives: To compare the predictive performance, generalization capability, and stability of multiple ML models for detecting elevated PSA levels via binary classification of PSA status.
Methods: We implemented three ML algorithms with two feature selection approaches to address the events-per-variable (EPV) problem. Lifestyle and demographic data were collected from adult males in Etsako West LGA, Edo State, Nigeria. Models were trained on 70% of the data (n=69) and validated on 30% (n=30). Performance was assessed using accuracy, precision, recall, specificity, F1-score, and ROC-AUC. Five-fold stratified cross-validation evaluated model stability. Hyperparameter optimization was performed using GridSearchCV.
Results: The Random Forest model achieved the most balanced performance with 73.3% accuracy, 70.6% precision, 80.0% recall, 63.6% specificity, 0.750 F1-score, and 0.764 ROC-AUC. SVM showed identical test set performance (73.3% accuracy, 0.764 ROC-AUC). Logistic Regression with L1 regularization and 11 features achieved the highest recall (100%) but at the cost of zero specificity, indicating overfitting. Cross-validation revealed model stability: Random Forest CV recall 0.833 ± 0.061, CV F1 0.813 ± 0.053. The F1-optimized Random Forest showed improved balance (70.0% accuracy, 66.7% recall, 50.0% specificity). All models demonstrated ROC-AUC between 0.714–0.764, indicating acceptable discrimination capability.
Conclusions: Random Forest and SVM demonstrated the most balanced performance in terms of sensitivity and specificity for PSA prediction in a small-sample setting. The study highlights important methodological considerations, including the need for feature selection under EPV constraints, the role of regularization in mitigating overfitting, and the importance of cross-validation for evaluating models out of sample performance.The moderate performance (ROC-AUC ≈ 0.71–0.76) suggests that lifestyle-based ML models may be useful for preliminary screening but are not suitable for diagnostic applications. Future work should focus on larger datasets, external validation, and ensemble methods.},
month = {March},
doi = {https://doi.org/10.64388/IREV8I9-1714356}
}