A Hybrid AI Approach: Integrating Fuzzy Logic, QSPR and Artificial Neural Networks for Heart Attack Drug Characterization

ABSTRACT Cardiovascular diseases, such as acute myocardial infarction, present ongoing global health challenges, necessitating advanced computational tools for drug evaluation and optimization. This study proposes a novel hybrid framework that integrates fuzzy artificial neural networks (FANN) with quantitative structure–property relationship (QSPR) modeling to accurately predict key physicochemical properties of anti‐cardiovascular drugs. By leveraging fuzzy logic, the model effectively captures uncertainty and imprecision inherent in chemical and biological data, while QSPR establishes rigorous quantitative links between molecular structure and properties such as density, boiling point, ACD LogP, and enthalpy of vaporization. A dataset comprising clinically relevant cardiovascular drugs, including antiplatelet agents, antiplatelet agents (clopidogrel and ticagrelor), beta‐blockers (metoprolol and carvedilol), and thrombolytics (alteplase), was used to train and evaluate the models. FANN consistently outperformed conventional models, including Random Forest, with superior predictive accuracy as reflected in low error metrics (MSE: 28.83, MAE: 3.39, RMSE: 5.37) and a high value approaching 0.99. The results demonstrate FANN's robustness in modeling nonlinear structure property relationships and its potential utility for virtual screening and early‐stage compound prioritization. This research contributes to computational cardiology by offering a reliable, interpretable, and cost‐effective strategy for predicting physicochemical properties of cardiovascular drugs.