Efficient feature selection with attention based deep cat convolutional stacked sparse autoencoder for diabetes...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Diabetes mellitus is a global health concern requiring early detection for effective management.
• Current diagnostic methods can be improved with advanced computational techniques for earlier identification.

Purpose of the Study:

• To develop and validate a novel deep learning model for the early detection of diabetes.
• To enhance feature selection for improved diagnostic accuracy in diabetes prediction.

Main Methods:

• Data preprocessing and feature selection using an improved Cheetah Optimization (ICO) algorithm.
• Classification of diabetes using a dual attention-based deep cat convolutional stacked sparse autoencoder (DA_DCC_SSAE) model.
• Evaluation of the model's performance on multiple datasets.

Main Results:

• The proposed DA_DCC_SSAE model achieved high accuracy rates across four datasets: 98.4% (dataset-1), 98% (dataset-2), 97.4% (dataset-3), and 96.8% (dataset-4).
• The ICO feature selection method contributed to improved classification performance.
• The deep learning approach demonstrated significant potential for early diabetes identification.

Conclusions:

• The novel deep learning framework, incorporating ICO and DA_DCC_SSAE, offers a promising and highly accurate method for early diabetes detection.
• This approach can aid in timely interventions, potentially preventing diabetes complications.
• Further research can explore the clinical integration of this AI-driven diagnostic tool.