Classification of Manifest Huntington Disease Using Vowel Distortion Measures

Abstract

Huntington disease (HD) is a fatal autosomal dominant neurocognitive disorder that causes cognitive disturbances, neuropsychiatric symptoms, and impaired motor abilities (e.g., gait, speech, voice). Due to its progressive nature, HD treatment requires ongoing clinical monitoring of symptoms. Individuals with the Huntingtin gene mutation, which causes HD, may exhibit a range of speech symptoms as they progress from premanifest to manifest HD. Speech-based passive monitoring has the potential to augment clinical information by more continuously tracking manifestation symptoms. Differentiating between premanifest and manifest HD is an important yet under-studied problem, as this distinction marks the need for increased treatment. In this work we present the first demonstration of how changes in speech can be measured to differentiate between premanifest and manifest HD. To do so, we focus on one speech symptom of HD: distorted vowels. We introduce a set of Filtered Vowel Distortion Measures (FVDM) which we extract from read speech. We show that FVDM, coupled with features from existing literature, can differentiate between premanifest and manifest HD with 80% accuracy.

Predicting clinical scores in Huntington’s disease: a lightweight speech test

“…In their study, rhythmic deficits correlated with the TMS only when measured from a reading task (Percentage of silence R = 0.4) and a monologue task (Percentage of silence R = 0.5) but not from automated speech (recitation of the days of the week, Percentage of silence R = 0.08). Similarly, although HD participants have difficulties to sustain the vowel /a/ steadily for a few seconds compared to premanifest patients [ 30 ], speech features extracted from this simple task could not improve the clinical score extracted from demographics alone [ 38 ]. In our present study, we used both an automatic task (counting forward) and a more cognitive complex task (counting backward).…”

“…Among these markers, it was found that the speech rate correlates with disease burden score, probability of disease onset, the estimated years to onset, and cognitive score [ 19 , 27 ]. In addition, speech analysis combined with machine learning models allowed the discrimination of manifest HD and PreHD individuals from controls [ 29 , 30 ]. However, some of these speech tasks suffer some drawbacks, such as the requirement of fastidious annotation by linguistic experts or language adaptation difficulties, which make their use not suitable for clinical practice; and their sensitivity to the various HD symptoms remain unknown [ 31 ].…”

A comparison study on patient-psychologist voice diarization

“…While works studying these problems in isolation has provided valuable insights, in this work, we showed that Speaker Role Recognition was the most suitable approach for Interviewees at different stages of Huntington's Disease. For future work, we plan to investigate the use of these methods to derive robust biomarkers automatically and compare them to more classic approaches Perez et al, 2018;Romana et al, 2020).…”

Feasibility of a continuous, multi-sensor remote health monitoring approach in persons living with neurodegenerative disease

“…The decision to include a chest-mounted device with both ECG and accelerometry capabilities was guided by evidence to suggest that continuous ECG outperforms proxy measures obtained via photoplethysmography (PPG) [ 45 ]. There was also an interest in examining autonomic function and other relevant measures derived from a sensor located on the chest, such as speech or respiration, which are known to be affected in persons with CVD and NDD [ 46 – 48 ].…”