Detecting Low Frequency Oscillations in Cardiovascular Signals Using Gradient Frequency Neural Networks

Gradient Frequency Neural Networks (GFNNs) have been applied successfully to detect pulse and meter (hierarchical groupings of pulses) in complex music audio signals having polyrhythms and syncopation. Here, we apply GFNNs to the detection of low frequency (LF) and high frequency (HF) oscillations in cardiovascular signals, namely the heart rate variations associated with Mayer waves and with respiration, respectively. The cardiovascular time series is treated as music audio for analysis; the electrocardiographic (ECG) signal is processed as a WAV file, and R-R intervals converted to a MIDI file. GFNNs are networks of nonlinear neural oscillators that offer the advantage of high sensitivity at low stimulus amplitudes, compared to linear amplitude responses, for weak signals. The GFNNs entrained with prominent LF peaks at 0. 0837 ± 0.0175Hz to R-R intervals of Kundalini meditators from the Physio Bank Exaggerated Heart Rate Oscillations database. When applied to a 15-hour Holter recording of Paroxysmal Atrial Fibrillation, GFNN entrainment showed significant LF activity between 0.04-0.14Hz, and HF activity at ~ 0.25Hz during sleep. GFNNs present a novel approach to the detection and study of cardiovascular oscillations, inspired by auditory rhythm perception.