A robust algorithm for estimating the balance of Autonomic Nervous System with application to mental fatigue detection using photoplethysmographic (PPG) signals
Abstract
Spectral- and time-domain analysis of Heart Rate Variability (HRV) signal is widely used as a quantitative marker of the Autonomic Nervous System (ANS) activity. A robust algorithm was developed to derive HRV from photoplethysmographic (PPG) signals, to compute FFT- and AR-based spectra of these HRV signals, and to determine time- and frequency-domain features. This algorithm has detrending, sample-rate reductions, false-peak removal, automatic peak detection, peak-to-peak (P-P) interval detection and correction, time-domain feature extraction, HRV signal generation, and spectral-domain feature extraction from the HRV signal. Adapting to the very low spectral contents of the input PPG signal is very helpful in reducing the processing/computational effort. The spectral features include the LF/HF ratio since this can be used to quantify parasympathetic influences and sympathovagal balance. To validate the efficacy of the algorithm, PPG signals were recorded under different conditions such as stimulating an acupuncture point using a nanoscale patch, measuring relaxation after exercising, and others which are known to elicit changes in the state of the ANS. Significant differences in LF/HF were observed due to these effects. The pNN50, a time-domain measure of PP interval variability, was also considered for quantifying ANS activity and exploring its correlation with spectral features. We also used multiple sensors placed on different fingers to record PPG signals and to confirm that their respective spectral analysis was almost identical. We observed that a multiple sensor approach could be used to effectively reduce the impact of motion artifacts and of deterioration of signal quality due to loss of good PPG sensor contact. Finally, a time varying approach for analysis of HRV signal spectra was developed. It is proposed as a tool to estimate the ANS balance at any particular instant of time.
Subject Area
Electrical engineering
Recommended Citation
Verma, Ajay Kumar, "A robust algorithm for estimating the balance of Autonomic Nervous System with application to mental fatigue detection using photoplethysmographic (PPG) signals" (2014). ETD Collection for University of Texas, El Paso. AAI1557801.
https://scholarworks.utep.edu/dissertations/AAI1557801