Content uploaded by Daksh Abrol
Author content
All content in this area was uploaded by Daksh Abrol on Feb 20, 2025
Content may be subject to copyright.
Fast Fourier Transform in Assessing the Effect of
Parkinson Disease Progression on Sympathetic
Oscillation
Daksh Abrol, Kumar Satyam, and Yogender Aggarwal
The complex relationship between sympathetic oscillation and the progression of Parkinson disease (PD) is still under active
investigation. This study examines the influence of PD on sympathetic activity by examining heart rate variability derived from
electrocardiogram recordings. The peak frequency within the low-frequency band of heart rate variability, which is primarily
associated with sympathetic modulation, was extracted and compared across various phases of PD severity. As PD ad-
vanced, the results revealed a significant and progressive decline in peak frequency, indicating a reduction in sympathetic re-
sponsiveness. This observation corresponds with the recognized autonomic dysfunction that accompanies PD progression.
Furthermore, a significant negative correlation was observed between peak frequency and disease severity, suggesting its
potential utility as a biomarker for monitoring PD advancement and treatment efficacy. The findings emphasize the clinical rel-
evance of autonomic dysfunction in PD and highlight the potential of peak frequency analysis as an effective method for
adding to the diagnosis, prognosis, and management of this complex neurodegenerative disorder.
Parkinson disease (PD) is a neurodegenerative condition pri-
marily characterized by motor symptoms. These motor manifesta-
tions come from the gradual loss of dopaminergic neurons in the
substantia nigra pars compacta, a brain area critical for movement
regulation.
1
However, PD also comprisesavarietyofnonmotor
symptoms, including autonomic dysfunction, which may signifi-
cantly impact patients’everyday life.
2
The autonomic dysfunction
might emerge as orthostatic hypotension,
3
constipation, urine dif-
ficulties, sweating abnormalities, and cardiac irregularities.
4
The balance between the sympathetic and parasympathetic
branches is crucial for maintaining homeostasis and adjusting
to internal and external stimuli.
The delicate connection between PD progression and sympa-
thetic oscillation remains an elusive element of this neurodegener-
ative sickness, requiring thorough examination. The typical motor
symptoms of PD, including bradykinesia, stiffness, tremor, and
postural instability, are commonly known; nevertheless, the
widespread effect of PD goes beyond motor dysfunction, cover-
ing a range of nonmotor manifestations, especially autonomic
dysfunction.
5,6
The autonomic nerve system, a complex network
managing key physiological activities such as heart rate, blood
pressure, and digestion, is severely disrupted by the neurodegen-
erative processes inherent in PD. This change in autonomic func-
tion may instigate a cascade of consequences, including cardio-
vascular irregularities, gastrointestinal abnormalities, and ther-
moregulatory deficits, significantly reducing the overall quality
of life for patients with PD.
Heart rate variability (HRV), a noninvasive figure derived from
electrocardiogram (ECG) recordings, gives a window into the con-
stantly changing relationship between the sympathetic and para-
sympathetic branches of the autonomic nerve system.
7
The peak
frequency within the low-frequency (LF) band of HRV, primarily
related to sympathetic modulation, has emerged as a viable indica-
tor for measuring autonomic function in PD.
8
Prior studies have
suggested abnormalities in HRV patterns in PD patients, suggest-
ing a shift in the delicate balance between sympathetic and para-
sympathetic activity.
The present work aimed to analyze ECG frequency with the se-
verity of PD. This work intended to understand the subtle link be-
tween disease development and sympathetic response. The find-
ings of this research hold the potential to contribute to knowledge
of the pathophysiological mechanisms underlying autonomic dys-
function in PD, paving the way for the development of innovative
diagnostic and prognostic tools.
Methodology
Study Design and Participants
A total of 156 subjects were enrolled in the study involving 92
PD and 64 healthy controls. The institutional ethics commit-
tee of Rajendra Institute of Medical Sciences, Ranchi, ap-
proved this study (No-05, dated January 29, 2020), and all partic-
ipants gave their written informed consent. All subjects underwent
Corresponding author: Yogender Aggarwal is an assistant professor in the
Department of Bioengineering and Biotechnology at Birla Institute of Technology in
Mesra, Ranchi, Jharkhand, India, and can be reached at yaggarwal@bitmesra.ac.in.
Daksh Abrol is a student in the Department of Electrical and Electronics
Engineering at the Birla Institute of Technology in Mesra, Ranchi, Jharkhand,
India, and can be reached at btech10022.21@bitmesra.ac.in.
Kumar Satyam is a Research Scholar in the Department of Bioengineering and
Biotechnology at the Birla Institute of Technology, in Mesra, Ranchi, Jharkhand,
India, and can be reached at phdbe10055.18@bitmesra.ac.in.
Journal of Clinical Engineering, (2025) 50, 1, 22–26
The authors declare no conflicts of interest.
http://dx.doi.org/10.1097/JCE. 0000000000000680
Copyright © 2025 Wolters Kluwer Health, Inc. All rights reserved.
Feature Article
22
Copyright © 2024 Wolters Kluwer Health, Inc. All rights reserved.
Downloaded from http://journals.lww.com/jcejournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC1y0abggQZXdtwnfKZBYtws= on 12/18/2024
clinical testing including the motor part of the Movement Disorder
Society’s Unified Parkinson Disease Rating Scale, a widely used
scaling technique for PD severity.
9,10
This scale consists of 65 items
in 4 parts with a total score of 260. Based on UPDRS scaling, this
research adopted a cross-sectional strategy, with varied degrees of
PD severity, divided into 4 unique cohorts:
•Healthy controls: UPDRS < 5
•Mild PD: UPDRS ≤65
•Moderate PD: 65 < UPDRS < 135
•Severe PD: 135 ≤UPDRS
The healthy control group comprised those without any
neurological abnormalities, acting as a baseline for compari-
son. The inclusion and exclusion criteria for each cohort were
carefully specified to reduce confounding variables and en-
sure the homogeneity of the groupings.
Data Acquisition and Preprocessing
High-quality ECG recordings were collected from each participant
utilizing a consistent methodology in a controlled setting. The
MP45 bioamplifier (Biopac Systems Inc, USA) was used to record
the lead II ECG for 10-minute duration. The ECG signals were
preprocessed to increase data quality and prepare them for further
analysis. The preprocessing processes included the following:
(1) Filtering: Bandpass filtering was done to eliminate
noise and artifacts beyond the frequency range of in-
terest (0.05-35 Hz). The MATLAB (The Mathworks,
Natick, MA, USA) code achieved this using a
Butterworth bandpass filter:
[b_bandpass, a_bandpass] = butter(filter_order, [fcl_bandpass
fch_bandpass]/(fs/2), ‘bandpass’);
filtered_ecg = filtfilt(b_bandpass, a_bandpass, data);
(2) Artifact removal: Advanced algorithms were ap-
plied to detect and eradicate artifacts such as muscle
noise, baseline drift, and powerline interference.
The MATLAB algorithm tackled powerline interfer-
ence using a notch filter:
[b_bandstop, a_bandstop] = butter(filter_order, [f_low_bandstop
f_high_bandstop]/(fs/2), ‘stop’);
filtered_ecg = filtfilt(b_bandstop, a_bandstop, filtered_ecg);
(3) Segmentation: Each ECG recording was segmented
into shorter segments (1-minute epochs) to ease analysis
across consistent periods and enable the extraction of ac-
curate frequency. The MATLAB code accomplished this
segmentation inside its main loop:
for minute = 1:adjusted_minutes start_index = (minute - 1) *
60*fs+1;
end_index = min(minute * 60 * fs, length(data)); segment =
filtered_ecg(start_index:end_index);
Signal Processing and Feature Extraction
The preprocessed ECG data received additional refinement via fil-
tering to improve them for further analysis. A high-pass filter with
a cutoff frequency of 2 Hz was used to minimize any residual
baseline drift or LF noise. The MATLAB code achieved this using
a Butterworth high-pass filter:
[b_highpass, a_highpass] = butter(filter_order, fc_highpass/
(fs/2), ‘high’);
filtered_ecg = filtfilt(b_highpass, a_highpass, data);
Each 1-minute section was thereafter submitted to frequency-
domain analysis utilizing the FAST Fourier transform (FFT) to ac-
quire its power spectrum. The peak frequency was subsequently
identified as the key indication of sympathetic modulation. This
peak frequency was retrieved for each segment and served as an im-
portant characteristic for statistical analysis. The power spectral
density of each ECG segment will be computed using the Welch
technique, with suitable windowing and overlap. The peak fre-
quency was designated as the primary result.
The following MATLAB code sample explains the essence
of this process:
N = 2^nextpow2(length(segment));
Y = fft(segment, N);
P1 = abs(Y/N);
P1 = P1(1:N/2 + 1);
P1(2:end-1) = 2*P1(2:end-1);
f = fs*(0:(N/2))/N;
[~, domFreqIndex] = max(P1(f < =50));
dominantFrequency = f(domFreqIndex);
Visualization and Output
The MATLAB function creates figures to view the ECG seg-
ments, FFT spectra, and power spectral density plots for each
minute of the recording. The “subplot”function was used to or-
ganize numerous plots inside a single figure. The “plot”function
was applied to construct line plots, whereas the “xlabel,”
“ylabel,”and “title”functions added labels and titles to
the plots. The figures were saved in both “.fig”and “.png”
formats for additional study and display.
Finally, the feature table containing the extracted peak fre-
quency was written to an Excel (Microsoft Corp, Redmond,
Washington) file (.xlsx) using the “writetable”function. This
provides simple access and additional analysis of the data
using spreadsheet software or other statistical tools.
Statistical Analysis
The obtained peak frequency in each group (healthy controls, mild
PD, moderate PD, severe PD) was tested statistically for significant
differences at P<.05.Further,thecorrelationanalysiswasper-
formed to explore the association between LF peak frequency
and PD severity.
Results
The peak frequency within the LF band of HRV, which is pre-
dominantly linked with sympathetic activity, was extracted
and compared across various stages of PD severity (Table,
Figures 1 to 4). Figures 1 to 4 demonstrated the identification
TABLE
The Average Peak Frequencies for Each Group
Serial No. Stages Average Peak Frequency, Hz
1Healthy controls 4.9170
2 Mild PD 4.6325
3 Moderate PD 4.3998
4 Severe PD 4.2038
23
www.jcejournal.com Volume 50 Issue 1 2025 Journal of Clinical Engineering
Copyright © 2024 Wolters Kluwer Health, Inc. All rights reserved.
Downloaded from http://journals.lww.com/jcejournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC1y0abggQZXdtwnfKZBYtws= on 12/18/2024
of peak frequency for healthy, mild, moderate, and severe
subjects using FFT methodology.
The gradual decrease in peak frequency was demonstrated
with the severity of the disease from mild to severe in compar-
ison to healthy subjects (Figure 5).
The Student ttest at P< .05 indicated that:
•There is a significant difference between healthy
and moderate.
•There is a significant difference between healthy
and severe cases.
•There is a significant difference between mild and
moderate cases.
•There is a significant difference between mild and
severe cases.
•There is a significant difference between the mod-
erate and the severe cases.
These data imply a steady drop in peak frequency as PD
evolves from moderate to severe stages. Furthermore, a sub-
stantial negative connection was detected between peak fre-
quency and disease severity, indicating its potential value as a
biomarker for monitoring PD development. The distribution of
FIGURE 2. Analysis of an ECG signal segment for a particular minute epoch of a patient with mild PD. Abbreviations: ECG, electrocardiogram; PD, Parkinson disease.
FIGURE 1. Analysis of an ECG signal segment for a particular minute epoch of a healthy patient. Abbreviation: ECG, electrocardiogram.
24
www.jcejournal.com Volume 50 Issue 1 2025 Journal of Clinical Engineering
Copyright © 2024 Wolters Kluwer Health, Inc. All rights reserved.
Downloaded from http://journals.lww.com/jcejournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC1y0abggQZXdtwnfKZBYtws= on 12/18/2024
peak frequencies throughout distinct PD stages and the connec-
tion with disease severity are graphically displayed.
Discussion
The present study aimed to analyze ECG signals to correlate
LF peak frequency and severity of PD. The obtained results
suggested that the ECG signal frequency gradually decreases
with the severity of the disease. The peak frequency within
the LF band is associated with sympathetic activity. Our find-
ings suggested that the reduction in frequency resulted in au-
tonomic dysfunction and is associated with the withdrawal
of the parasympathetic activity.
11
The observed findings sug-
gested that cardiac parasympathetic modulation in PD re-
sulted from confounding factors.
12
The value of LF reduced
significantly more in the PD group than in the control group.
Consistent with previous reports, we found that PD patients
older than 60 years had reduced frequency with disease sever-
ity levels compared with healthy controls suggesting auto-
nomic and baroreflex dysfunction.
13,14
The review of the lit-
erature revealed that time-domain HRV parameters do not
reliably reflect PD progression.
15
Previous studies described
the differences between patients and healthy controls with re-
spect to the HRV parameters.
16,17
FIGURE 4. Analysis of an ECG signal segment for a particular minute epoch of a patient with mild PD. Abbreviations: ECG, electrocardiogram; PD, Parkinson disease.
FIGURE 3. Analysis of an ECG signal segment for a particular minute epoch of a patient with moderate PD. Abbreviations: ECG, electrocardiogram; PD,
Parkinson disease.
25
www.jcejournal.com Volume 50 Issue 1 2025 Journal of Clinical Engineering
Copyright © 2024 Wolters Kluwer Health, Inc. All rights reserved.
Downloaded from http://journals.lww.com/jcejournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC1y0abggQZXdtwnfKZBYtws= on 12/18/2024
The 3 frequency-domain measures LF, HF, and LF/HF and
UPDRS motor score showed the most consistent associations
between autonomic parameters and clinical scales.
15
Our re-
sults are in line with this finding, which suggests a significant
correlation between peak frequency LF for sympathetic re-
sponse and the disease severity based on UPDRS scaling.
Parkinson disease results in sympathovagal imbalance and is
correlated with disease severity.
Conclusion
This research into the influence of PD development on sympa-
thetic oscillation has revealed intriguing findings. The ob-
served drop in peak frequency within the LF band is sugges-
tive of the substantial involvement of autonomic dysfunction
in PD. The significant association between peak frequency
and disease severity further increases its usefulness as a nonin-
vasive biomarker for monitoring PD progression and therapy
response. The results not only add more to the delicate link
between PD and autonomic function but also open the way
for the creation of novel diagnostic and prognostic tools.
References
[1] Elsheikh S, Coles NP, Achadu OJ, Filippou PS, Khundakar AA.
Advancing brain research through surface-enhanced raman
spectroscopy (SERS): current applications and future prospects.
Biosensors. 2024;14:33.
[2] Kaufmann H, Norcliffe-Kaufmann L, Palma JA. Baroreflex
dysfunction. N Engl J Med. 2020;382:163-178.
[3] Francois C, Biaggioni I, Shibao C, et al. Fall-related healthcare use
and costs in neurogenic orthostatic hypotension with Parkinson’s
disease. J Med Econ. 2017;20:525-532.
[4] Sabino-Carvalho JL, Falquetto B, Takakura AC, Vianna LC.
Baroreflex dysfunction in Parkinson’s disease: integration of central
and peripheral mechanisms. JNeurophysiol. 2021;125:1425-1439.
[5] Titova N, Qamar MA, Chaudhuri KR. The nonmotor features of
Parkinson's disease. Int Rev Neurobiol. 2017;132:33-54.
[6] Heimrich KG, Lehmann T, Schlattmann P, Prell T. Heart rate
variability analyses in Parkinson’s disease: a systematic review and
meta-analysis. Brain Sci. 2021;11:959.
[7] Kamath MV, Watanabe M, Upton A. Heart Rate Variability (HRV)
Signal Analysis: Clinical Applications. Boca Raton: CRC Press
Books; 2016.
[8] You S, Kim HA, Lee H. Association of postural instability with
autonomic dysfunction in early Parkinson’s disease. JClinMed.
2020;9:3786.
[9] El Maachi I, Bilodeau GA, Bouachir W. Deep 1D-Convnet for
accurate Parkinson disease detection and severity prediction from
gait. Expert Syst Appl. 2020;143:113075.
[10] Prashanth R, Roy SD. Novel and improved stage estimation in
Parkinson's disease using clinical scales and machine learning.
Neurocomputing. 2018;305:78-103.
[11] Yoon JH, Kim MS, Lee SM, Kim HJ, Hong JM. Heart rate
variability to differentiate essential tremor from early-stage tremor-
dominant Parkinson's disease. J Neurol Sci. 2016;368:55-58.
[12] Naranjo CC, Marras C, Visanji NP, et al. Short-term deceleration
capacity of heart rate: a sensitive marker of cardiac autonomic
dysfunction in idiopathic Parkinson’s disease. Clin Auton Res.
2021;31:729-736.
[13] Kallio M, Haapaniemi T, Turkka J, et al. Heart rate variability in
patients with untreated Parkinson’s disease. Eur J Neurol. 2002;7:
667-672.
[14] Barbic F, Perego F, Canesi M, et al. Early abnormalities of vascular
and cardiac autonomic control in Parkinson’s disease without
orthostatic hypotension. Hypertension. 2007;49:1 20-126.
[15] Maetzler W, Karam M, Berger MF, et al. Time-and frequency-
domain parameters of heart rate variability and sympathetic skin
response in Parkinson’s disease. J Neural Transm. 2015;122:
419-425.
[16] Reimann M, Friedrich C, Gasch J, Reichmann H, Rüdiger H,
Ziemssen T. Trigonometric regressive spectral analysis reliably
maps dynamic changes in baroreflex sensitivity and autonomic
tone: the effect of gender and age. PLoS One. 2010;5:e12187.
[17] Reimann M, Schmidt C, Herting B, et al. Comprehensive autonomic
assessment does not differentiate between Parkinson’s disease,
multiple system atrophy and progressive supranuclear palsy.
JNeuralTransm. 2010;117:69-76.
How to cite this article: Abrol D, Satyam K, Aggarwal Y. Fast fourier
transform in assessing the effect of parkinson disease progression on
sympathetic oscillation. JClinEng2025;50(1):22–26. doi: 10.1097/
JCE.0000000000000680
FIGURE 5. Plot of frequency over various PD stages with trend line. Abbreviation: PD, Parkinson disease.
26
www.jcejournal.com Volume 50 Issue 1 2025 Journal of Clinical Engineering
Copyright © 2024 Wolters Kluwer Health, Inc. All rights reserved.
Downloaded from http://journals.lww.com/jcejournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC1y0abggQZXdtwnfKZBYtws= on 12/18/2024
