Implementation and Analysis of Classification Algorithms for Diabetes

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Computational Biology

Background:

• Diabetes is a prevalent global health concern requiring accessible and sustainable management.
• Effective and timely diagnosis is crucial for managing diabetes and preventing complications.

Purpose of the Study:

• To develop and evaluate a novel classification technique for early and accurate diabetes detection.
• To improve the efficiency and performance of diabetes diagnostic models.

Main Methods:

• A localized diabetes dataset was compiled.
• Several classification techniques (RBF NN, MLP NN, NBs, J48graft DT) were applied.
• Genetic Algorithm (GA) was used for attribute selection, reducing 12 attributes to 6.

Main Results:

• Comparative analysis showed that GA significantly enhances Receiver Operating Characteristic (ROC) and accuracy.
• GA effectively removes insignificant attributes, reducing computational cost and time.

Conclusions:

• The proposed strategy, utilizing GA for attribute selection, improves diabetes classification accuracy and efficiency.
• This approach demonstrates potential for application in diagnosing other medical conditions.