Medical diagnosis involves analyzing symptoms, test results, and patient histories, but uncertainty from vague symptoms and incomplete records complicates the process. Fuzzy logic-based systems address this issue but often depend on manual rule creation, which is time-consuming. This research proposes a hybrid approach integrating fuzzy logic with deep learning techniques (FL-DLT) for intelligent diagnosis. The framework combines adaptive neuro-fuzzy inference system (ANFIS) for handling uncertainty with convolutional neural networks (CNNs) for extracting features from medical images like X-rays and MRIs. ANFIS models relationships between symptoms, results, and diagnoses, while CNNs analyze medical images. Experimental results show high accuracy and reliability, even with noisy or incomplete data. The proposed approach can improve diagnostic accuracy and efficiency, supporting clinicians in decision-making. Key contributions include the development of the FL-DLT framework and its evaluation using a large dataset of patient records and medical images. Additionally, the research offers insights into the application of fuzzy logic and deep learning in medical diagnosis, highlighting their potential to enhance diagnostic outcomes and efficiency in clinical practice.

Medical diagnosis has become increasingly difficult, requiring sophisticated systems to manage decision-making uncertainty. The proposed medical diagnosis utilizing fuzzy logic framework (MD-FLF) addresses medical data imprecision and ambiguity by employing fuzzy inference techniques. Clinical data ambiguity is often overlooked by conventional diagnosis models. Such models include probabilistic classifiers and threshold-based decision trees. The models use accurate input-output correlations. However, fuzzy inference simplifies incremental membership assignment and rule-based reasoning in MD-FLF. This enhances the system’s diagnostic ambiguity detection. The framework uses fuzzy rules to represent complicated non-linear interactions between symptoms and diagnostic data. This improves framework interpretation. MD-FLF models’ ambiguity and non-linear relationships between diagnostic inputs and outputs provide interpretable recommendations for complex disorders. Rule-based methods and expert knowledge produce these results. Experimental evaluations showed that MD-FLF improved reliability by 97.68%, uncertainty by 96.84%, ambiguity by 43.56%, patient variability by 98.26%, and diagnostic accuracy by 97.82%. The paradigm addresses uncertainty to increase diagnostic reliability, precision, and confidence while eliminating ambiguity and offering clinical decision-making insights. MD-FLF outperforms deterministic techniques in medical diagnostic decision support systems and is stable and interpretable.