Objective: To utilize free speech characteristics to develop markers of motor and cognitive status in Parkinson’s disease (PD).

Background: Vocal biomarkers based on speech volume, pitch, rate, articulation, and other characteristics have recently been used in detection of a variety of neurological disorders including PD. In our previous work, we have introduced novel markers based on the specific characteristics of articulatory coordination and phoneme-based pitch and rate in PD. We used these features successfully in detecting PD motor symptoms from structured (e.g., read) speech [1]. We now expand this work to free speech, which has not been well studied in PD. Free speech requires multiple motor and cognitive processes working in parallel to maintain fluency, and may be impacted by early cognitive changes. We therefore assessed free speech in non-demented PD patients to develop markers of cognitive and motor function. 

Methods: Speech recordings of description of the cookie theft picture in 35 non-demented (by consensus evaluation) PD subjects were analyzed using automated signal processing software. Speech features characterizing articulatory coordination based on temporal dynamics of resonant (formant) frequencies and delta-mel cepstral coefficients (dMFCC), as well as phoneme-dependent speaking rates, were analyzed. Using cross-validation methodology, Gaussian staircase regression models [1] were used to predict motor severity (MDS-UPDRS) and global cognition (MoCA). Area under the curve (AUC) in receiver operating characteristic (ROC) curves were used to quantify detection of MDS-UPDRS and MoCA scores.

Results: Prediction models for MDS-UPDRS and MoCA were based on four formant, five dMFCC, and four phoneme rate features including silent pauses. Using cutpoints of MDS-UPDRS < or ≥ 17 and MoCA < or ≥ 28, the following classification results were obtained: AUC = 0.84 for MDS-UPDRS detection, and AUC = 0.81 for MoCA detection.

Conclusions: Free speech markers estimated motor severity and global cognition in PD. Given these results in non-demented PD patients, speech markers may have potential to detect early cognitive changes. Speech markers could be implemented on a large scale through remote technology and may be a useful outcome measure in PD clinical trials. Speech markers therefore warrant further study in additional patient cohorts.

References: [1] Williamson, J. R., Quatieri, T. F., Helfer, B. S., Perricone, J., Ghosh, S. S., Ciccarelli, G., & Mehta, D. D. (2015, September). Segment-dependent dynamics in predicting Parkinson’s disease. In Sixteenth Annual Conference of the International Speech Communication Association.

« Back to 2017 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/speech-markers-estimate-motor-severity-and-global-cognition-in-parkinsons-disease/