Integrating Neural, Fuzzy, and Bio-Inspired Paradigms for Applications in Hypertension Modeling

Abstract

This work proposes a hybrid neuro-fuzzy model to provide accurate and timely diagnosis of the risk of developing hypertension and cardiovascular events. It utilizes artificial neural networks, in both modular and monolithic forms, as well as fuzzy classifiers. Each component of the model has been optimized using various bio-inspired algorithms to maximize both the accuracy and robustness of the system. The proposed modules include blood pressure behavior analysis, classifiers for nocturnal blood pressure profiles, heart rate, blood pressure level and load, as well as assessment of the risk of developing hypertension and identification of cardiovascular events. The results indicate that risk predictions achieved accuracy rates of up to 100%.