A secondary cognitive task can increase resting tremors in PD patients [13, 14]. Therefore, we developed a methodology to examine force control and maximum force on one hand while quantifying tremors on the other hand. Participants performed a 40-second isometric constant grip force task at 20% of their maximum voluntary contraction with their less affected hand for the first 20 seconds while receiving visual feedback on the force. For the subsequent 20 seconds, visual feedback was removed. Patients with no tremors used their dominant hand. We collected grip force data and acceleration data from the contralateral hand that was resting in a natural posture. Our aim was to investigate the force control and tremor characteristics of PD patients with different tremor profiles and severity levels. We recruited Parkinson's disease (PD) patients with unilateral, bilateral, and no tremors to take part in our study. The present study follows the Guidelines and Regulatory Standards for research involving human subjects established by the National Council of Health, Ministry of Health, in Brazil in October 1996, and was approved by the Ethics Committee of Anhembi Morumbi University, São Paulo, Brazil, (protocol # 3903038).
The population was composed of patients diagnosed with Parkinson's Disease, and the inclusion criteria included subjects who: a) had a diagnosis of PD; b) aged between 60 and 75 years; c) Hoehn & Yahr staging 2 and 3; d) had adequate comprehension skills and were able to carry out the proposed activities; e) did not have other associated pathologies; f) were in the "On" phase of medication. Exclusion criteria were subjects who: a) presented with orthopedic deformities; b) had epilepsy or other neurological diseases; c) had any type of metal in the head, surgical clip, welding fragment, firearm projectile, metal plates; d) had implanted devices such as a pacemaker, Deep Brain Stimulation (DBS), or cochlear implant; e) needed to use gait aids; f) had a freezing of gait episode (> 15 points) according to the Freezing of Gait Questionnaire [15].
Anthropometric data collection and measurement (weight, height, body mass index - BMI), Part III (motor) of the Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS) were obtained in the pre-intervention evaluation. MDS-UPDRS was used to determine tremor characteristics, as well as the most affected side and severity [16]. A cut-off point was found between mild/moderate and moderate/severe levels at 32/33 and 58/59, respectively. All subjects were taking Prolopa 200/50 and Levodopa.
Hand grip strength
Force data were collected using a manual grip dynamometer (Vernier Software & Technology, USA). Subjects were standing in front of a 14-inch LCD monitor that displayed a graph and held the dynamometer with their dominant hand. They were then required to exert the maximum force possible with one of the upper limbs generating force in finger flexion (palmar grip) and maintaining that force for 3 seconds. Three force spikes were performed for each hand of everyone, and an average of the spikes was calculated.
A maximal voluntary contraction (MVC) was quantified as the average force over a minimum of one second of the constant force of the highest attempt.
For the constant force control test, the subjects were required to sustain the palm grip position for 40 seconds, keeping a constant force of 20% of the constant MVC. During the first 20 seconds, the subjects received visual feedback of the force produced by the hand and the target force, while, in the following 20 seconds, the subjects continued to perform the constant force but without visual feedback [17, 18].
Accelerometry
A triaxial accelerometer model 3D-BTA, Vernier, with 30mA @5 VDC, ± 49 m/s2 range (± 5 g), the precision of ± 0.5 m/s2 (± 0.05 g), frequency response: 0-100 Hz and resolution: 0.037 m/s² was placed on the contralateral hand to the hand holding the dynamometer with the help of elastic bandages. The accelerometer was connected to Vernier Graphical Analysis software installed on a computer, and data was collected while the participant performed the dynamometry procedure.
Data Analyses
To analyze our data, we estimated force variability as the standard deviation and coefficient of variation of the detrended force in the time domain, and force error as the root mean square error. Force signals were band-pass filtered between 0.05 Hz and 10 Hz (Butterworth, order 4) and detrended Matlab 7.0.1 (MathWorks Inc.).
The frequency domain analysis was done using Morlet wavelet transform. Wavelet, cross-wavelet (x-wav) and wavelet coherence (WCS) spectra were generated using Matlab 7.0.1 (MathWorks Inc.) functions developed by several research groups [19–21]. All estimation were done considering 4 different frequency bands, 0-0.5 Hz, 0.5-3 Hz, 3–7 Hz, and 7–12 Hz [22–24]. First, force and acceleration wavelet absolute power spectra were estimated as the squared weighted modulus of the wavelet transform normalized by scale (Eq. 1,[25]).
$$WPS(s,\tau {)}^{X}=\frac{{\left|{W}^{X}(s,\tau )\right|}^{2}}{s}$$
1
where s represents the dilation parameter (scale shifting), τ represents the location parameter (time shifting) and Wx represents the complex wavelet transform of signal X.
Second, to measure the relative importance of different frequency bands of force and acceleration signals, we defined the normalized wavelet scale-averaged power spectrum (NWPS, Fig. 2) as the squared weighted modulus of the wavelet transform normalized by the sum of the squared weighted modulus over all scales for each instant to time (Eq. 2; [26, 27]).
The normalized wavelet scale-averaged power spectrum shows the relative importance through time of the frequency content of a signal with intensities ranging from 0-100%.
Additionally, we defined the cross-wavelet scale-averaged power spectrum (XWPS) as the weighted modulus of the cross-wavelet transform normalized over all scales (Eq. 3; [26, 28]).
$$XWPS(s,\tau {)}^{XY}=\frac{\left|{W}^{XY}(s,\tau )\right|}{s}$$
3
where Wxy represents the complex cross wavelet transform of signals X and Y. The cross-wavelet scale-averaged power spectrum shows the combined importance through time of the commonalities in the frequency content of two signals.
Finally, we quantified the wavelet coherence spectrum (WCS, [29]) was quantified as localized correlation coefficients in the wavelet time-frequency space as follows (Eq. 4):
$$WCS(s,\tau {)}^{XY}=\frac{{\left|S\left({s}^{-1}{W}^{XY}\left(s,\tau \right)\right)\right|}^{2}}{S\left({s}^{-1}{\left|{W}^{X}(s,\tau )\right|}^{2}\right).S\left({s}^{-1}{\left|{W}^{Y}(s,\tau )\right|}^{2}\right)}$$
4
where S is naturally designed as a smoothing operator with a similar footprint as the Morlet wavelet [20]. The wavelet coherence spectrum shows the correlation between the common frequencies of two signals with values ranging from 0–1.
Statistics
Regarding our time domain variables, we used a general linear model (GLM) with repeated measures, including one within-factor of “Condition” (with or without feedback) and two between factors of “Groups” (unilateral, bilateral, or no tremor) and “Severity” (mild or moderate). For our frequency domain variables, we performed another GLM with repeated measures including two within factors of “Condition” (with or without feedback) and “Frequency Bands” (0-0.5, 0.5-3, 3–7, 7–16 Hz), as well as the same between factors mentioned above. Moreover, in the context of our force wavelet power spectra analysis, we conducted another GLM, replacing “Groups” with a “Tremor” factor (tremor vs no tremor), which considers the presence or absence of tremor in the hand performing the force task.
All analyses were conducted using SPSS (version 20.0) statistical software (SPSS, Inc., Chicago, Illinois). Significant interactions found in the GLM models were further examined using appropriate post-hoc analyses. Main effects and significant interactions were followed by suitable comparisons using analyses of variance (ANOVA) of minor factors and dependents and independent t-tests. Multiple t-test comparisons were corrected with the Bonferroni correction. The statistical significance level for all tests was set at 0.05. The GLM results, reported as Fisher’s test (F) results, were followed by corresponding p-values and Partial Eta Squared (ηp2). The text shows data presented as mean ± SD, while the figures represent data as mean ± standard error of the mean (SEM). Unless otherwise stated, we present only significant main effects and interactions.