Joint generalized models for multidimensional outcomes: a case study of neuroscience data from multimodalities

Area of Science:

• Neuroscience
• Biostatistics
• Statistical Modeling

Background:

• Muscle fatigue research involves complex, multidimensional data from various sources.
• Simultaneous data acquisition from handgrip force, electromyography (EMG), and functional magnetic resonance imaging (fMRI) presents unique analytical challenges.

Purpose of the Study:

• To develop and evaluate statistical methods for analyzing multidimensional neuroscience data in muscle fatigue studies.
• To enable joint modeling of outcomes from different modalities to understand covariate effects and response associations.

Main Methods:

• Individual modeling of univariate responses using mixed-effects beta, simplex, and negative-binomial models for force/EMG percentages and fMRI counts.
• Development of a joint modeling approach to simultaneously analyze multidimensional outcomes from multiple modalities.
• Simulation studies to assess the performance of the proposed methods in finite sample situations.

Main Results:

• The proposed joint modeling approach allows for estimation of covariate effects across different data types.
• The methods facilitate evaluation of the strength of association among multiple responses from EMG, fMRI, and force measurements.
• Simulation results demonstrate the benefits of the new approaches for analyzing neuroscience data.

Conclusions:

• The developed statistical framework provides a robust method for analyzing complex, multidimensional neuroscience data in muscle fatigue research.
• Joint modeling enhances the understanding of neural and muscular mechanisms underlying muscle fatigue by integrating diverse data sources.