Comprehensive, Population-Based Sensitivity Analysis of a Two-Mass Vocal Fold Model

Area of Science:

• Acoustics
• Biomechanics
• Speech Science

Background:

• Previous vocal fold modeling focused on single subjects or limited configurations.
• A gap exists in understanding population-level model behavior and parameter influence.

Purpose of the Study:

• To conduct a broad, population-based sensitivity analysis of a vocal fold model.
• To identify trends and influential relationships between model parameters and outputs across a virtual population.
• To inform optimization of reduced-order models for voice abnormality research.

Main Methods:

• Employed four distinct sensitivity analysis techniques.
• Examined a virtual population of subjects to identify inter-individual trends.
• Explored the parameter space of a selected two-mass vocal fold model.

Main Results:

• Vocal fold model behavior is dominated by complex interactions between parameters.
• Few input-output relationships demonstrated consistent effects across the virtual population.
• Identified influential relationships between specific model input parameters and outputs.

Conclusions:

• Sensitivity analysis on vocal fold models reveals complex, non-linear dynamics.
• Findings aid in optimizing reduced-order models for voice disorder research.
• Highlights challenges and provides recommendations for future complex vocal fold model sensitivity analyses.