A Medical Expert System for the Diagnosis of Cardiovascular Disease using Integrated DI-LSTM and T-ANFIS Rule-Based Models
1
Research Scholar, School of Computer and Systems Sciences, Jaipur National University, Jaipur, Rajasthan, India.
2
Professor, HOD School of Computer and Systems Sciences, Jaipur National University, Jaipur, Rajasthan, India
Received: 2025-07-14
Revised: 2025-08-22
Accepted: 2025-09-19
Published: 2025-10-04
| This research presents a deep learning framework for the diagnosis of heart disease. The proposed model combines a Dropout-layer Induced Long Short-Term Memory (DI-LSTM) network with a Trapezoidal-based Adaptive Neuro-Fuzzy Inference System (T-ANFIS).The objective of this combination is to improve the computational efficiency and the diagnostic accuracy. Firstly, the ECG signals pre-processed and decomposed by removing baseline drift and smoothing. Then the most relevant features such as statistical properties, time-related factors, amplitude measurements, and pulse characteristics are extracted. We optimize these features with a Dent Map-based Red Fox Optimization (TM-RFO) algorithm, which improves search capabilities and avoids getting stuck in local minima by using Dent Map dynamics. The improved feature set is fed into a DI-LSTM framework. Dropout layers are used to improve generalization and prevent overfitting. This allows for binary classification of signals as normal or abnormal. If signals are abnormal, they go through a secondary classification using the T-ANFIS framework. To handle complex data patterns and identify conditions such as atrial fibrillation (AFib), ventricular fibrillation (VFib), and bradycardia, we use structural adaptive trapezoidal membership functions. Our framework shows that it performs better in terms of sensitivity, specificity, recall, accuracy, and precision metrics compared to existing methods.
Dropout-layer Induced Long Short-Term Memory, Tent Map-based Red Fox Optimization, Trapezoidal Adaptive Neuro-Fuzzy Inference System, deep neural network, Cardiac Disease Diagnosis.