An Integrated Approach for Diabetes Detection Using Fisher Score Feature Selection and Capsule Network

This study explores the use of Fisher score feature selection combined with capsule networks for diabetes detection, achieving high performance with an accuracy, precision, recall, and F1 score all at 94%. The approach effectively captures relevant information for accurate classification, demonstrating its potential as a reliable tool for diagnosing diabetes by correctly identifying both positive and negative cases, thus reducing misdiagnosis risk.