Functional link convolutional neural network for the classification of diabetes mellitus

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Diabetes mellitus is a metabolic disorder characterized by hyperglycemia due to defects in insulin action, secretion, or both.
• Intelligent systems are increasingly utilized for disease classification, with deep learning showing significant promise.
• Existing machine learning techniques for diabetes classification vary in complexity and performance.

Purpose of the Study:

• To propose and evaluate a Functional Link Convolutional Neural Network (FLCNN) for diabetes classification.
• To assess the potential of a computationally less complex deep learning network for medical diagnosis.
• To compare the performance of FLCNN against other machine learning techniques using a real-world diabetes dataset.

Main Methods:

• A variant of Convolutional Neural Network, the Functional Link Convolutional Neural Network (FLCNN), was developed for diabetes classification.
• The proposed FLCNN model was applied to a real dataset collected from Bombay Medical Hall, Ranchi, India.
• Comparative analysis was conducted by implementing and evaluating various other machine learning techniques alongside FLCNN.

Main Results:

• The proposed FLCNN classifier achieved an accuracy exceeding 90% in diabetes classification.
• Performance metrics were evaluated using standard measures and validated with the non-parametric Friedman test.
• The FLCNN demonstrated competitive accuracy compared to other state-of-the-art implemented classifiers.

Conclusions:

• The Functional Link Convolutional Neural Network (FLCNN) presents a viable and effective deep learning approach for diabetes classification.
• FLCNN offers a computationally efficient alternative for disease classification tasks in healthcare.
• The study validates the potential of FLCNN for accurate and reliable diabetes diagnosis using real-world data.