Objective: We characterized complex language profile and explore diagnostic potential of automated language analysis of natural spontaneous speech in differentiation between multiple system atrophy (MSA) and Parkinson’s disease (PD).

Background: Patients with synucleinopathies such as MSA and PD frequently display language abnormalities, however detailed language profile in MSA has never been investigated.

Method: A total of 39 participants with MSA compared to 39 age- and sex-matched de-novo untreated PD, and 39 age- and sex-matched healthy controls were recruited. The PD group was also matched to MSA by disease duration (mean 4.2, SD 1.6 years), estimated based on the self-reported manifestation of the first motor symptoms. All participants were guided to perform monologue on a topic of their preference lasting approximately 2 minutes. Each audio sample was transcribed and linguistically annotated using automatic speech recognition and natural language processing. A quantitative analysis was performed using 6 lexical and syntactic and 2 acoustic features. Results were compared with human-controlled analysis to assess robustness of the approach. Diagnostic accuracy was evaluated using sensitivity analysis.

Results: Despite similar disease duration, language abnormalities were generally more severe in MSA than in PD, leading to high diagnostic accuracy with an area under the curve of 0.81 (sensitivity 77%; specificity 74%). Compared to controls, MSA showed decreased grammatical component usage (p<0.001), more repetitive phrases (p=0.002), shorter sentences (p<0.001), reduced sentence development (p<0.001), slower articulation rate (p<0.001), and increased duration of pauses (p<0.001), whereas PD had only shorter sentences (p<0.001), reduced sentence development (p=0.003), and longer pauses (p=0.002). Only slower articulation rate was distinct for MSA while unchanged for PD. Despite relatively high severity of dysarthria in MSA, a strong reliability between manually and automatically computed results was achieved (Pearson’s r = 0.60-0.94).

Conclusion: Integrating automated language analysis could provide a cost-effective, time-efficient, and widely applicable novel tool to estimate the extent of cognitive impairment and early detect synucleinopathies with similar clinical manifestations.

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/characteristics-of-language-abnormalities-in-multiple-system-atrophy-and-parkinsons-disease/