Differentiating upper- and lower motor neuron diseases using automated acoustic analysis.

OBJECTIVE

Motor neuron diseases (MNDs) result in a spectrum of motor impairments, including considerable effects on speech function, which manifest as dysarthria-a motor speech disorder. Speech metrics are increasingly recognized as critical biomarkers with potential utility in disease diagnosis and phenotyping. This study aimed to (1) characterize acoustics of upper motor neuron (UMN) and lower motor neuron (LMN) dysarthria presentations in MNDs, and (2) identify relationships between bulbar disease severity scores and acoustic features, as these could collectively enable personalized approaches to management of these diseases.

METHODS

Data from 16 individuals with primary lateral sclerosis (PLS) representing UMN disease, 14 individuals with spinal and bulbar muscular atrophy (SBMA) representing LMN disease, and 25 neurologically healthy individuals were analyzed. Clinical measures were also collected from PLS and SBMA groups. All participants were remotely recorded performing passage reading, rapid syllable repetition, and vowel phonation. Fifty-two acoustic features were extracted representing articulation, phonation, prosody, resonance, and overall speech timing. Features were compared using Kruskal-Wallis tests for between-group comparisons and Spearman correlations between acoustic features and clinical scores.

RESULTS

Articulatory and prosodic features best differentiated PLS, SBMA and controls. Correlations were observed in the PLS group between the clinical score and various articulatory features, most notably those indexing tongue and jaw movements.

CONCLUSIONS

Our study demonstrated that acoustic assessment could capture fingerprints of dysarthrias associated with PLS and SBMA. These findings also demonstrate the potential for remote speech assessment to characterize diverse dysarthria profiles and pave the way for creating ways for personalized disease management approaches in clinical care and trials.