Using Digital Speech Markers to Classify Functional Speech Disorder: A Proof-of-Concept Pilot Study

Functional speech disorder (FND-speech) is a subtype of functional neurological disorder, yet quantitative characterization of its motor and cognitive-linguistic features remains underexplored.

This study aimed to quantitatively characterize FND-speech by comparing acoustic and linguistic features in individuals with FND-speech with healthy control subjects (HCs). The potential of digital speech features to serve as adjunctive diagnostic markers was also evaluated.

Thirty adults with FND-speech (females, n = 25; males, n = 5) and 47 age-, sex-, education-, and handedness-matched HCs (females, n = 29; males, n = 18) were compared using lexicosyntactic, rate-based, and acoustic markers extracted from a structured picture description task. Supervised machine learning was employed to classify FND-speech versus HCs.

Lexicosyntactic features showed moderate predictive power (area under the curve [AUC] = 0.80), as did rate-based features (AUC = 0.81). Acoustic features demonstrated high discrimination (AUC = 0.98), and a combined model incorporating all feature categories achieved similar performance (AUC = 0.98).

This proof-of-principle pilot study successfully classified FND-speech versus HCs, highlighting the potential of digital speech markers to serve as adjunctive diagnostic markers for FND-speech. Out-of-sample replication and larger-scale classifier efforts incorporating neurological controls are needed. © 2025 International Parkinson and Movement Disorder Society.