Fuzzy Logic and CNN-Based Hybrid Models in Medical Decision Support: A Review of Interpretable and Uncertainty- Aware Artificial Intelligence

Artificial intelligence is increasingly transforming medical decision support by enabling rapid analysis of complex clinical data. Convolutional Neural Networks have shown strong performance in image-based and signal-based medical tasks, particularly because of their ability to learn hierarchical representations from raw data. However, their limited interpretability and weak explicit handling of uncertainty remain major barriers to clinical adoption. Fuzzy logic, in contrast, provides a transparent reasoning framework that can model vague clinical concepts such as severity, instability, urgency, and risk through linguistic rules. This review discusses the potential of combining fuzzy logic and CNN-based deep learning to build medical decision support systems that are both accurate and explainable. Evidence from emergency triage, AI-based triage systems, multimodal emergency decision support, and wound classification suggests that hybrid fuzzy-CNN models can improve decision consistency, uncertainty management, and multimodal data integration. Nevertheless, challenges such as external validation, workflow integration, rule complexity, data bias, and real-world clinical deployment remain unresolved. This review argues that fuzzy-CNN hybrid systems represent a promising direction for future clinical AI, particularly in high-risk environments where interpretability and uncertainty-aware decision-making are essential.

کلیدواژه ها:

Fuzzy logic ، CNN ، medical decision support ، emergency triage ، wound classification