Fig 1.
Representative voice recordings and spectrograms of SLE patients and healthy controls.
A: Sound recordings of /a/ vowel phonations from a control subject (A1) and a patient with SLE (A2); note the high variability of the signal from the SLE patient in relation to the control. B: Zoomed in views of the recordings presented in A1 and A2; there is a noticeable reduction in the amplitude range in the SLE patient voical signal in relation to the representative control recording. C: Spectrogram (frequency domain) representations of the recordings presented in A1 and A2; note the high level of background noise and low formant segregation in the SLE patient compared to the healthy control.
Fig 2.
Objective and subjective (GRBAS) vocal parameters of control subjects (n = 32) and SLE patients (n = 36).
Horizontal lines represent the population median. Each graph shows values for healthy controls and SLE patients of A: F0; B: vocal intensity; C: jitter (main formant frequency variability); D: shimmer (intensity variability); E: HNR; F: G (general grade of dysphonia); G: R (roughness); H: B (breathiness); I: A (asthenia); J: S (strain); * = P < 0.05; ** = P < 0.001. *** = P < 0.0001.
Fig 3.
Linear regressions between objective and subjective vocal parameters with potential determinants of dysphonia and self-reported vocal deficits.
A: Matrix representing the R2 values of linear regressions between selected variables. Note the high correlations (R2 > 0.15) between HNR, G, B, R and S with the number of self-reported vocal deficits. B: Matrix representing the P values of linear regressions between selected variables. Note the significant correlations (P < 0.05) between HNR, G, B, R and S with the number of self-reported vocal deficits. C: Scatter-plot representations of the significant correlations identified in the correlation matrices in A and B.
Fig 4.
Linear regressions between objective and subjective vocal parameters with tissue damage as measured by the SLICC/ACR damage index.
A: Matrix representing the R2 values of linear regressions between selected variables. Note the high correlations (R2 > 0.15) between intensity with summed damage scores and renal, cardiovascular, musculoskeletal and skin scores, as well as between pulmonary damage scores with jitter, shimmer and HNR. B: Matrix representing the P values of linear regressions between selected variables. Note the significant correlations (P < 0.05) between intensity with summed damage scores and renal, cardiovascular, musculoskeletal and skin scores, as well as between pulmonary damage scores with jitter, shimmer and HNR. C: Scatter-plot representations of selected significant correlations identified in the correlation matrices in A and B.