A Comprehensive Review and Future Directions in Diabetes Prediction

Abstract

Accurate and timely diagnosis of diabetes is critical for improving patient outcomes. This paper presents a comprehensive review of the application of the Adaptive Neuro-Fuzzy Inference System (ANFIS) in medical image classification, with a specific focus on diabetes prediction utilizing retinal fundus images and optical coherence tomography (OCT). Leveraging the synergistic capabilities of fuzzy logic and neural networks, ANFIS emerges as a promising tool for handling the complexities of medical data, particularly in tasks related to diabetic retinopathy and macular edema detection. The review explores ANFIS's effectiveness, emphasizing its interpretability, adaptability to uncertain data, and capacity to model nonlinear relationships. However, the challenge of parameter tuning is acknowledged, prompting suggestions for future research directions. The integration of deep learning techniques is proposed to enhance ANFIS's performance, addressing the evolving demands of medical image classification. The insights provided aim to guide researchers toward refining ANFIS models and advancing automated diagnostic tools for diabetes prediction.

Keywords

References