A New Multi-Agent Feature Wrapper Machine Learning Approach for Heart Disease Diagnosis

Heart disease (HD) is a serious widespread life-threatening disease. The heart of patients with HD fails to pump sufficient amounts blood the entire body. Diagnosing occurrence early and efficiently may prevent manifestation debilitating effects this aid in its effective treatment. Classical methods for diagnosing are sometimes unreliable insufficient analyzing related symptoms. As an alternative, noninvasive medical procedures based on machine learning (ML) provide reliable diagnosis efficient prediction conditions. However, existing models automated ML-based diagnostic cannot satisfy clinical evaluation criteria because their inability recognize anomalies extracted symptoms represented as classification features from HD. In study, we propose (AHDD) system that integrates binary convolutional neural network (CNN) new multi-agent feature wrapper (MAFW) model. MAFW model consists four software agents operate genetic algorithm (GA), support vector (SVM), Naïve Bayes (NB). instruct GA perform global search adjust weights SVM BN during initial classification. A final tuning CNN then performed ensure best set included identification. five layers categorize healthy or according analysis optimized features. We evaluate performance proposed AHDD via 12 common ML techniques conventional by using cross-validation technique assessing six criteria. achieves highest accuracy 90.1%, whereas other attain only 72.3%–83.8% average. Therefore, herein has capability identify This can be used practitioners diagnose efficiently.