Using fuzzy association rule mining in cancer classification.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning in Oncology

Background:

• Gene expression profiling is crucial for cancer tumor classification.
• Existing methods often lack model interpretability.
• Identifying significant genes and their relationships is key to understanding disease outcomes.

Purpose of the Study:

• To develop an interpretable fuzzy classifier for cancer tumors.
• To achieve high classification performance using selected significant genes.
• To integrate gene selection and fuzzy rule generation.

Main Methods:

• Utilized fuzzy rules and linguistic variables for model interpretability.
• Employed various gene selection methods to identify significant gene subsets.
• Developed a new algorithm based on fuzzy association rule mining to generate fuzzy rules and select genes.
• Combined genes associated in primary classifiers to create a novel classifier.

Main Results:

• The fuzzy classifier demonstrated high performance in tumor classification.
• The model successfully presented relationships between genes using linguistic variables.
• The integrated approach effectively selected significant genes and generated interpretable rules.

Conclusions:

• Interpretable fuzzy classifiers can achieve high performance in cancer gene expression analysis.
• Fuzzy association rule mining provides a robust framework for gene selection and rule generation.
• This approach offers a promising tool for understanding gene-disease relationships in oncology.