Remote smartphone-based spoken language screening predicts clinical markers in Huntington's disease

Area of Science:

• Neurology
• Computational Linguistics
• Digital Health

Background:

• Current Huntington's disease (HD) symptom assessment relies on infrequent clinical evaluations.
• Remote monitoring of HD progression is needed for efficient clinical trials and patient care.

Purpose of the Study:

• To evaluate the efficacy of automated linguistic analysis of smartphone-recorded speech tasks for capturing core clinical markers of HD severity.
• To determine if digital biomarkers derived from speech can remotely quantify HD impairment.

Main Methods:

• A cross-sectional multicenter study involving 53 participants (30 with HD, 23 healthy controls).
• Participants completed 7 days of smartphone-based speech tasks (monologue, fairy tale retelling).
• Natural language processing analyzed speech for lexical and syntactic features to predict clinical outcomes (Unified Huntington's Disease Rating Scale, cognitive scales).

Main Results:

• Linguistic features significantly predicted HD severity, explaining up to 63% of the variance in motor impairment.
• HD participants exhibited reduced vocabulary range and increased phrase repetition compared to controls.
• Automated speech analysis using smartphones provides a scalable, low-burden digital biomarker for HD.

Conclusions:

• Fully automated analysis of smartphone-based language assessments can remotely quantify cognitive, motor, and functional impairments in Huntington's disease.
• This approach offers a promising digital biomarker for decentralized monitoring and clinical trials in HD.
• Speech analysis via smartphone enables objective, frequent assessment of HD progression.