Dermatology Disease Prediction Based on Two Step Cascade Genetic Algorithm Optimization of ANFIS Parameters

Area of Science:

• Dermatology
• Artificial Intelligence
• Machine Learning

Background:

• Accurate dermatological disease prediction is crucial for effective treatment.
• Existing models often struggle with real-world clinical data variability.
• Need for robust algorithms utilizing only clinical features and diagnoses.

Purpose of the Study:

• To develop and present novel algorithms for dermatological disease prediction.
• To optimize fuzzy models using a combination of Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Genetic Algorithms (GA).
• To validate the model's performance using real-world clinical data.

Main Methods:

• Utilized Adaptive Neuro-Fuzzy Inference Systems (ANFIS) for fuzzy model creation.
• Employed Genetic Algorithms (GA) for optimizing ANFIS subtractive clustering parameters (first level).
• Applied a second level of GA optimization on the genetically optimized ANFIS structure.
• Implemented double 2-fold cross-validation for robust model assessment in MATLAB.

Main Results:

• The proposed ANFIS-GA hybrid model demonstrated enhanced prediction accuracy.
• Achieved higher accuracy rates compared to previously established models.
• Validated effectiveness using real-world dermatological clinical features and diagnoses.

Conclusions:

• The novel ANFIS-GA algorithms offer a significant improvement in dermatological disease prediction.
• The model's ability to perform with real-world data highlights its practical applicability.
• This approach provides a more accurate and reliable tool for clinical decision support.