Effect of fuzzy partitioning in Crohn's disease classification: a neuro-fuzzy-based approach

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Gastroenterology

Background:

• Crohn's disease (CD) diagnosis poses significant challenges due to its severe gastrointestinal impact and complex treatment needs.
• Accurate and early diagnosis of CD is crucial for effective patient management and improving outcomes.
• Existing diagnostic methods may have limitations in sensitivity, specificity, or efficiency.

Purpose of the Study:

• To develop and evaluate a hybrid neuro-fuzzy model for enhanced Crohn's disease diagnosis.
• To investigate the impact of factor analysis for data dimension reduction on diagnostic performance.
• To determine the optimal fuzzy partitioning level for maximizing classification accuracy.

Main Methods:

• Utilized a backpropagation neural network fuzzy classifier integrated into a neuro-fuzzy model.
• Employed factor analysis for dimensionality reduction of the diagnostic dataset.
• Systematically varied fuzzy partitioning levels to assess their effect on model performance.
• Evaluated the system using classification accuracy, sensitivity, and specificity.

Main Results:

• The hybrid neuro-fuzzy model demonstrated high diagnostic performance.
• Factor analysis and fuzzy partitioning were found to influence system accuracy.
• Level-8 fuzzy partitioning achieved the highest classification accuracy of 97.67%.
• The optimal model achieved 96.07% sensitivity and 100% specificity.

Conclusions:

• The proposed neuro-fuzzy model offers a promising approach for accurate Crohn's disease diagnosis.
• Optimized data dimension reduction and fuzzy partitioning are key to achieving high diagnostic performance.
• This AI-driven method has the potential to aid clinicians in diagnosing CD more effectively.