Relationship Between Voice Analysis and Functional Status in Patients with Amyotrophic Lateral Sclerosis

Abstract

Background: Amyotrophic Lateral Sclerosis (ALS) is a progressive neurodegenerative disease affecting both upper and lower motor neurons, with bulbar dysfunction manifesting in up to 80% of patients. Dysarthria, characterized by impaired speech production, is common in ALS and often correlates with disease severity. Voice analysis has emerged as a promising tool for detecting disease progression and monitoring functional status. Methods: This study investigates acoustic and biomechanical voice alterations in ALS patients and their association with clinical measures of functional independence. A descriptive observational case series study was conducted, involving 43 ALS patients and 43 age and sex matched controls with non-neurological voice disorders. Sustained vowel /a/ recordings were obtained and analyzed using Voice Clinical Systems^® and Praat software (version 6.2.22). Biomechanical and acoustic parameters were correlated with ALS Functional Rating Scale-Revised (ALSFRS-R) and Barthel Index scores. Results: Significant differences were observed between ALS and control groups (elevated muscle force and tension and interedge distance in non-ALS individuals). Between bulbar and spinal ALS subtypes, elevated values were observed in certain parameters in Bulbar ALS patients, indicating irregular vocal fold contact and weakened phonatory control, while spinal ALS exhibited increased values, suggesting higher phonatory muscle tension. Elevated biomechanical parameters were significantly correlated with low ALSFRS-R scores, suggesting a possible relationship between voice measures and functional decline. However, acoustic measurements showed no relationship with performance status. Conclusions: These results highlight the potential of voice analysis as a non-invasive, objective tool for monitoring ALS stage and differentiating between subtypes. Further research is needed to validate these findings and explore their clinical applications.

Keywords: Amyotrophic Lateral Sclerosis; acoustic analysis; dysarthria; functionality.

Figure 1

This schematic illustrates the flowchart employed in the study of acoustic and biomechanical changes in the voice of ALS patients, in relation to their functional status.

Figure 2

Differences in vocal parameters between ALS and non-ALS individuals. Loadings (bar plots) represent the contribution of each variable to the discriminant components and may vary across models depending on the data and components analyzed: (a) Bar plot showing differences in vocal parameters between bulbar ALS patients and control individuals. Warm colors indicate significant R² values. Variables whose bars are higher in the positive Y-axis region are increased in the bulbar ALS group. Conversely, variables whose bars are higher in the negative Y-axis region are higher in the control. (b) Magnified view of the selected parameters from (a), emphasizing those with higher R² values. In the OPLS-DA bar plots, the Y-axis represents the regression coefficients (weights) assigned to each vocal parameter by the model. These coefficients indicate the relative contribution of each parameter to the discrimination between groups (e.g., ALS subtypes vs. controls). Higher absolute values suggest a stronger influence of a given parameter on the separation, with positive and negative values reflecting differences in feature expression between the compared groups. Since OPLS-DA is a supervised multivariate method, the original vocal parameters are transformed into a latent space that maximizes inter-group variance while minimizing intra-group variance. The regression coefficients in the bar plots result from this transformation and provide insight into which features most effectively distinguish ALS subtypes from controls. Importantly, the scale of the Y-axis is model-dependent and varies according to the magnitude of the coefficients derived from each specific comparison. (c) Violin plot for Pr12, the only parameter showing a statistically significant difference in bulbar ALS patients compared to controls (t-test, p-value = 3.08 × 10⁻⁵). The boxplot within the violin plot represents the data distribution and central tendencies. (d) Bar plot comparing spinal ALS patients to control individuals, following the same color scheme as (a). Pr21 is higher in controls.

Figure 3

Differences in vocal parameters between bulbar ALS and spinal ALS patients: Bar plot showing the differences in vocal parameters between bulbar and spinal ALS patients based on an O-PLS DA model (R² = 0.17). Warm colors indicate significant R² values according to the O-PLS DA model. Values that are significantly higher in the bulbar ALS are plotted in the positive Y-axis region. Higher values in spinal ALS are plotted in the negative Y-axis region. Three parameters (Pr5, Pr9, and Pr18) demonstrated significant differences between the two groups and are highlighted with boxplots representing their distributions. Boxplots show the normalized value of each parameter in bulbar ALS (pink) and spinal ALS (green), with yellow dots indicating mean values. Pr5 and Pr18 were significantly higher in bulbar ALS, whereas Pr9 was significantly higher in spinal ALS (p-values = 0.047, 0.02, and 0.045, respectively, after FDR correction).

Figure 4

ROC curve analysis and classification performance for ALS subtype differentiation based on voice parameters: (a–f) Classical ROC curve analysis for individual voice biomarkers (Pr5, Pr6, Pr7, Pr9, Pr18, Pr20). The ROC curves illustrate the true positive rate (sensitivity) versus the false positive rate (1-specificity) for each parameter. The AUC (area under the curve) values indicate the discriminative ability of each biomarker, with confidence intervals shown in parentheses. The red dot marks the optimal cutoff point based on the Youden Index. (g–l) Boxplots comparing selected voice biomarkers between bulbar ALS (pink) and spinal ALS (green) subtypes. The horizontal red line represents the median, while the yellow dot indicates the mean value for each group. (m) ROC curve of the O-PLS-DA classification model using the most discriminative features (Pr18, Pr9, Pr5). To generate a smooth curve, 100 cross-validations were performed, and the results were averaged. The shaded area represents the 95% confidence interval for the ROC curve, with an overall AUC of 0.71 (95% CI: 0.452–0.895). (n) Scatter plot of predicted class probabilities for bulbar and spinal ALS samples. The confusion matrix (inset) shows the classification results from cross-validation, indicating the number of correctly and incorrectly classified samples for each ALS subtype. (o) Boxplot of the prediction accuracy across cross-validation iterations. The median and interquartile range of model performance are displayed, highlighting variability in classification accuracy.

Figure 5

Differences in vocal parameters associated with the Barthel Index in ALS patients and control individuals: Bar plot illustrating the differences in vocal parameters between ALS patients and control individuals based on an O-PLS DA model (R² = 0.30). Warm colors indicate significant differences in the parameter of interest. When the bar is increased in the positive region of the Y axis, it means that the value is higher in the ALS group. Conversely, when the bar is increased in the negative region of the Y axis, the parameter is increased in the control group. A zoomed-in section highlights specific parameters with notable differences. Pr15 was significantly higher in ALS patients (in warm, intense color).

Figure 6

Differences in vocal parameters associated with ALSFRS-R scores in ALS patients and control individuals: Bar plot illustrating the differences in vocal parameters between ALS patients and control individuals based on an O-PLS DA model (R² = 0.30). As with the analysis presented in Figure 5, the OPLS-DA model used for Figure 6 was designed to assess the relationship between vocal parameters and ALSFRS-R scores within the ALS group, rather than to classify ALS versus control individuals. Significant R² values are colored in warm colors (i.e., red to yellow). The direction of the bars in Figure 6 does not indicate a positive or negative correlation but rather represents the measured vocal parameter values. Parameters plotted in the positive Y-axis region correspond to ALS patients, while those plotted in the negative Y-axis region correspond to controls, as indicated.

Figure 7

Influence of smoking and sex on vocal parameters in ALS patients: (A) O-PLS DA model assessing the effect of smoking on vocal parameters (R²Y = 0.20). The model suggests alterations in Pr20 levels, with smokers exhibiting significantly higher levels. (B) O-PLS DA model evaluating sex-related differences in vocal parameters (R²Y = 0.60). Significant differences were observed only for F₀ and Pr1. The warm color bars indicate the R²Y values are significant, with positive values representing higher levels in the women’s group.