Prediction of Skin Disease Using Ensemble Data Mining Techniques and Feature Selection Method-a Comparative Study.

Area of Science:

• Dermatology
• Machine Learning
• Artificial Intelligence

Background:

• Skin diseases pose a significant global health challenge.
• Accurate and timely diagnosis is crucial for effective treatment.
• Machine learning offers potential for automated skin disease classification.

Purpose of the Study:

• To evaluate the efficacy of ensemble machine learning techniques for skin disease prediction.
• To compare the performance of ensemble methods with individual classifiers and feature-selected models.
• To identify key features contributing to accurate dermatological diagnosis.

Main Methods:

• Applied six distinct machine learning algorithms.
• Developed an ensemble approach using bagging, AdaBoost, and gradient boosting.
• Implemented a feature importance method to select the top 15 predictive features.
• Compared model performance on the full dataset versus a feature-selected subset.

Main Results:

• Ensemble methods demonstrated superior prediction accuracy compared to individual classifiers.
• The ensemble approach outperformed models using only the feature-selected subset.
• Feature selection identified crucial indicators for skin disease prediction.
• Increased dermatological prediction accuracy was observed with the proposed ensemble method.

Conclusions:

• Ensemble machine learning techniques significantly enhance skin disease prediction accuracy.
• Combining ensemble methods with feature selection offers a robust approach for dermatological diagnosis.
• The proposed method provides a more accurate and effective solution for identifying skin disease classes.