A Decision Tree-Initialised Neuro-fuzzy Approach for Clinical Decision Support

Area of Science:

• Artificial Intelligence in Medicine
• Computational Intelligence
• Machine Learning for Healthcare

Background:

• Healthcare demands interpretable machine learning for validating medical knowledge.
• Fuzzy rule-based systems offer interpretability through linguistic if-then statements.
• Fuzzy sets handle vague medical concepts and enable approximate human-like reasoning.

Purpose of the Study:

• To develop a data-driven approach for learning accurate and interpretable fuzzy rule bases.
• To enhance clinical decision support systems with explainable artificial intelligence.
• To integrate fuzzy logic with adaptive network-based fuzzy inference systems (ANFIS) for optimized performance.

Main Methods:

• Generating an initial crisp rule base using decision tree learning.
• Transforming the crisp rules into a fuzzy rule base.
• Utilizing the Adaptive Network-based Fuzzy Inference System (ANFIS) for parameter optimization.

Main Results:

• The proposed method learns compact fuzzy rule bases with simple antecedents.
• Achieved statistically comparable or superior performance against state-of-the-art fuzzy classifiers.
• Demonstrated effectiveness on popular medical data benchmarks.

Conclusions:

• The approach successfully generates accurate and interpretable fuzzy rule bases for clinical decision support.
• It provides a valuable tool for clinicians to understand and validate AI-driven medical insights.
• This work advances the integration of explainable AI in healthcare applications.