سامانه همایش‌ها TSTA | ثبت‌نام و ارسال مقاله کنفرانس‌ها و کنگره‌های ملی

صفحه اصلی / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Prediction of cardiac arrhythmia via an improved hierarchical fused fuzzy deep learning

نویسندگان :

Arman Daliri¹ Nora Mahdavi²

1- دانشگاه آزاد اسلامی واحد کرج 2- دانشگاه آزاد اسلامی واحد کرج

کلمات کلیدی :

Cardiac arrhythmia،Deep Neural Networks،Non-sinusoidal arrhythmias،Fuzzy Deep Learning،Cardiovascular diagnostics

چکیده :

Deep Neural Networks, are widely applied in various fields such as healthcare. Cardiac arrhythmias, particularly non-sinusoidal arrhythmias, represent a critical challenge in cardiovascular diagnostics. As the volume of data generated by humans, including numbers, images, signals, and sounds continues to grow, the importance of DNNs in data analysis becomes more evident. This research focuses on cardiac arrhythmia classification using a Fuzzy Deep Neural Network (FDNN). The proposed FDNN integrates fuzzy logic and deep learning to effectively address traditional deep learning models’ limitations in handling data uncertainty and variability. FDNN demonstrates superior performance, achieving high accuracy, precision, and recall, supported by an impressive ROC curve and AUC value of 0.95, outperforming both traditional and advanced neural network-based methods. The interpretability of the FDNN model, supported by techniques like SHAP values, integrated gradients, and calibration analysis, further reinforces its potential for clinical adoption and trust in its decision-making process. The study highlights the potential of combining fuzzy logic with deep learning for medical applications, particularly in arrhythmia Prediction, offering a robust solution for accurate classification. Extended cross-validation (5- to 30-fold) confirms its robustness, with deep learning models showing consistent gains (~1–5%) over conventional approaches. These results highlight FDNN's potential for reliable arrhythmia detection, emphasizing the advantage of advanced architectures in medical diagnostic.