Related Experiment Video
Updated: Jan 9, 2026
13:19
Microarray-based Identification of Individual HERV Loci Expression: Application to Biomarker Discovery in Prostate Cancer
Published on: November 2, 2013
17.0K
Identification of Potential Biomarkers in Prostate Cancer Microarray Gene Expression Leveraging Explainable Machine
Ahmed Al Marouf1, Jon George Rokne1, Reda Alhajj1,2,3
1Department of Computer Science, University of Calgary, Calgary, AB T2N 1N4, Canada.
Cancers
|December 11, 2025
Summary
This study introduces Explainable Machine Learning (XML) to identify prostate cancer biomarkers. The novel approach achieved 81.01% accuracy using Random Forest, pinpointing key genes for personalized oncology.
Area of Science:
- Bioinformatics and Computational Biology
- Oncology
- Machine Learning
Background:
- Prostate cancer diagnosis and severity stratification are crucial for personalized treatment.
- Traditional biomarker discovery lacks transparency, hindering clinical trust.
- Bioinformatics offers tools, but interpretability remains a challenge.
Purpose of the Study:
- To develop and validate an Explainable Machine Learning (XML) approach for identifying and prioritizing prostate cancer biomarkers.
- To enhance the interpretability of machine learning models in bioinformatics for clinical decision-making.
- To discover severity-specific gene biomarkers for improved prostate cancer management.
Main Methods:
- Implemented various machine learning classifiers (Naive Bayes, Random Forest, Decision Tree, SVM, Logistic Regression, Bagging).
- Utilized SHAP (SHapley Additive explanations) values for model explainability.
- Applied data pre-processing techniques including imputation, SMOTE, and Tomek links for class imbalance, alongside k-fold stratified validation.
Main Results:
- Achieved a highest accuracy of 81.01% using the Random Forest model.
- Identified ten potential gene biomarkers: DEGS1, HPN, ERG, CFD, TMPRSS2, PDLIM5, XBP1, AJAP1, NPM1, and C7.
- Validated the model on a tissue microarray dataset of 102 patients.
Conclusions:
- Explainable Machine Learning (XML) effectively identifies severity-specific biomarkers in prostate cancer.
- This approach supports precision oncology by enabling targeted interventions.
- The findings herald a new era of individualized care for prostate cancer patients.
