Related Experiment Video
Updated: Jan 16, 2026
08:05
Detection and Isolation of Cancer in Prostate Biopsies Using Stimulated Raman Histology and Artificial Intelligence
Published on: June 10, 2025
1.1K
A simple and fast explainable artificial intelligence-based pre-screening tool for breast cancer tumor malignancy
Selahaddin Batuhan Akben1, Hilal Yumrutaş2
1Department of Electrical and Electronics Engineering, Osmaniye Korkut Ata University, Osmaniye, Türkiye. batu130@hotmail.com.
Scientific Reports
|October 2, 2025
Summary
This study introduces an explainable AI tool for fast breast cancer malignancy pre-screening. The decision tree model achieves high accuracy with minimal data, offering practical clinical decision support.
Area of Science:
- Oncology
- Artificial Intelligence
- Medical Informatics
Background:
- Early and accurate detection of breast cancer malignancy is critical for effective patient management.
- Current diagnostic methods may require extensive data and computational resources.
- There is a need for efficient and interpretable tools for breast cancer pre-screening.
Purpose of the Study:
- To develop and evaluate an explainable artificial intelligence (XAI)-based pre-screening tool for breast cancer malignancy.
- To assess the tool's performance using minimal clinical and demographic data.
- To ensure the tool is fast, accurate, and clinically practical.
Main Methods:
- Utilized a Kaggle dataset comprising 9 clinical and demographic features from 213 breast cancer patients.
- Compared the performance of 8 machine learning algorithms, including ensemble models and individual decision trees.
- Evaluated models based on accuracy, sensitivity, specificity, F1 score, AUC, and Matthews correlation coefficient.
- Employed SHapley Additive exPlanations (SHAP) for model interpretability.
Main Results:
- Both RUSBoost ensemble models and individual decision trees achieved approximately 91.7% accuracy.
- The decision tree model was chosen for its superior explainability, low computational cost, and clinical practicality.
- The decision tree model generated clear verbal rules for malignancy classification, considering factors like lymph node involvement, metastasis, tumor size, and patient age.
- SHAP analysis confirmed the model's decision-making process.
Conclusions:
- An XAI-based decision tree model offers a rapid, reliable, and low-data-requirement solution for breast cancer malignancy pre-screening.
- The model's explainability and clinical practicality make it suitable for integration into clinical decision support systems.
- Further validation studies with larger, diverse populations are warranted to improve generalizability.
