Cardiac Sudden Death Risk Detection Using Hybrid Neuronal-Fuzzy Networks

The cardiovascular diseases are the main cause of mortality in the industrialized world. The efforts to improve the diagnosis and the therapy in our days are highly developed. A noninvasive technique is the analysis of HRV (Heart Rate Variability) often from electrocardiography records (ECG) of 24 hours. HRV is the measurement of the interval between R peaks of two consecutive QRS complexes (RR intervals). An adaptive filter is used to eliminate the noise signals from muscular origin and signals from movements of the electrodes on the skin. Finally, power spectral density (PSD) are computed, filtering the signals in the three bands that characterize the HRV: high frequencies (HF), low frequencies (LF) and the very low frequencies (VLF). The model includes inputs from the time and frequency domain. We propose the application of combined neuronal networks with fuzzy logic systems that allow the quantification and characterization of the HRV, helping the identification of patients with low and high probability (risk) of undergoing a cardiac problem. The training procedure, its parameters and details of the application have been developed. The results suggest that this kind of hybrid network is suitable for the identification of patients with high/low cardiac risk. The simulation environment can be considered as a powerful tool for development methods in biomedical engineering particularly in cardiology