A novel muscle network approach for objective assessment and profiling of bulbar involvement in ALS.

Introduction: As a hallmark feature of amyotrophic lateral sclerosis (ALS), bulbar involvement significantly impacts psychosocial, emotional, and physical health. A validated objective marker is however lacking to characterize and phenotype bulbar involvement, positing a major barrier to early detection, progress monitoring, and tailored care. This study aimed to bridge this gap by constructing a multiplex functional mandibular muscle network to provide a novel objective measurement tool of bulbar involvement.

Methods: A noninvasive electrophysiological technique-surface electromyography-was combined with graph network analysis to extract 48 features measuring the regulatory mechanisms, connectivity, integration, segregation, assortativity, and lateralization of the functional muscle network during a speech task. These features were clustered into 10 interpretable latent factors. To evaluate the utility of the muscle network as a bulbar measurement tool, a heterogenous ALS cohort, consisting of eight individuals with overt clinical bulbar symptoms and seven without, along with 10 neurologically healthy controls, was employed to train and validate statistical and machine learning algorithms to assess the disease effects on the network features and the relation of the network performance to the current clinical diagnostic standard and behavioral patterns of bulbar involvement.

Results: Significant disease effects were found on most network features. The most robust effects were manifested by reduced and more variable myoelectric activities, and reduced functional connectivity and integration of the muscle network. The 10 latent factors (1) demonstrated acceptably high efficacy for detecting bulbar neuromuscular changes across all clinically confirmed symptomatic cases and clinically silent prodromal cases (area under the curve = 0.89-0.91; F1 score = 0.85-0.87; precision = 0.84-0.86; recall = 0.87-0.88); and (2) selectively correlated with clinically meaningful behavioral patterns (conditional = 0.45-0.81).

Conclusion: The functional muscle network shows promise for an objective quantifiable measurement tool to improve early detection and profiling of bulbar involvement across the prodromal and symptomatic stages. This tool has various strengths, including the use of a clinically readily available noninvasive instrument, fully automated data processing and analytics, and generation of interpretable objective outcome measures (i.e., latent factors), together rendering it highly scalable in routine clinical practice for assessing and monitoring of bulbar involvement.