Atrial fibrillation, characterized by chaotic rhythms and electrical complexity, presents a diagnostic challenge that requires innovative approaches to uncover its underlying biomarkers. This study proposes a hybrid predictive model based on multinomial logistic regression and neutrosophic logic, aiming to identify clinically significant patterns associated with this condition. Using the Knowledge Discovery in Databases (KDD) methodology, large volumes of cardiovascular data are analyzed to distinguish meaningful signals from background noise, revealing hidden connections and validating medical hypotheses. The implementation of the model through a digital prototype reflects a convergence of advanced statistics, artificial intelligence, and cardiovascular medicine, promoting a multidisciplinary approach. The findings of this work not only enhance diagnostic accuracy but also open new avenues for personalized treatment, emphasizing the value of scientific integration in modern medical research.