Breast Cancer Screening Based on Supervised Learning and Multi-Criteria Decision-Making.

Area of Science:

• Oncology
• Artificial Intelligence
• Machine Learning

Background:

• Early breast cancer detection significantly improves patient survival rates.
• Robust evaluation of machine learning models is crucial for reliable diagnostic tools.
• Current methods for model selection can be subjective and lack comprehensive analysis.

Purpose of the Study:

• To develop and apply a novel approach combining artificial intelligence (AI) and multi-criteria decision-making (MCDM) for evaluating machine learning (ML) models.
• To identify the most effective ML model for early breast cancer detection using the proposed integrated methodology.
• To provide a framework for robust model selection in medical diagnostics.

Main Methods:

• Employed supervised learning algorithms including Support Vector Machine, K-nearest neighbor, logistic regression, random forest, and naive Bayes classifiers.
• Utilized the Preference Ranking Organization Method for Enrichment Evaluations (PROMETHEE) as the MCDM technique.
• Calculated the net outranking flow for each ML model to determine their performance and ranking.

Main Results:

• The Support Vector Machine model achieved the highest net outranking flow (0.1022), indicating it as the most favorable model for early breast cancer detection.
• K-nearest neighbor, logistic regression, and random forest classifiers were ranked second, third, and fourth, respectively.
• The naive Bayes classifier was ranked as the least preferred model with a net outranking flow of -0.0766.

Conclusions:

• The proposed AI and MCDM integrated approach provides a robust and desirable method for selecting the optimal machine learning model for early breast cancer detection.
• This methodology allows for the incorporation of multiple criteria, enhancing decision-making in the selection of diagnostic AI tools.
• The findings highlight the potential of advanced computational methods to improve accuracy and efficiency in cancer diagnostics.