X-Vectors: New Quantitative Biomarkers for Early Parkinson's Disease Detection From Speech

Abstract

Many articles have used voice analysis to detect Parkinson's disease (PD), but few have focused on the early stages of the disease and the gender effect. In this article, we have adapted the latest speaker recognition system, called x-vectors, in order to detect PD at an early stage using voice analysis. X-vectors are embeddings extracted from Deep Neural Networks (DNNs), which provide robust speaker representations and improve speaker recognition when large amounts of training data are used. Our goal was to assess whether, in the context of early PD detection, this technique would outperform the more standard classifier MFCC-GMM (Mel-Frequency Cepstral Coefficients—Gaussian Mixture Model) and, if so, under which conditions. We recorded 221 French speakers (recently diagnosed PD subjects and healthy controls) with a high-quality microphone and via the telephone network. Men and women were analyzed separately in order to have more precise models and to assess a possible gender effect. Several experimental and methodological aspects were tested in order to analyze their impacts on classification performance. We assessed the impact of the audio segment durations, data augmentation, type of dataset used for the neural network training, kind of speech tasks, and back-end analyses. X-vectors technique provided better classification performances than MFCC-GMM for the text-independent tasks, and seemed to be particularly suited for the early detection of PD in women (7–15% improvement). This result was observed for both recording types (high-quality microphone and telephone).

A Multicenter Study Using a Smartwatch, Smartphone, and Wearable Sensors to Assess Early Parkinson’s Disease: Baseline Results of the WATCH-PD Study

“…This study, like others, also found differences by sex including in speech, and shows the value of connected speech tasks. 25,26 More research is needed, including longitudinal analysis from this study, to determine which measures are most sensitive to change over time.…”

Leveraging the Potential of Digital Technology for Better Individualized Treatment of Parkinson's Disease

“…Recent research shows that it is also possible to automatically discriminate PD from healthy controls at an early stage via handcrafted features extracted from speech ( 16 ). Moreover, in ( 17 ) a new DM technique (named X-vectors) was presented, which employs modern machine learning (ML) methods rather than designing handcrafted features. Improved classification results with X-vectors were obtained, validating the approach for early PD detection, using both high-quality speech recordings performed in laboratory settings and speech data acquired through telephone recordings.…”

Voice characteristics from isolated rapid eye movement sleep behavior disorder to early Parkinson's disease

“…First, we performed classifications from a subset of the high-level speech features, using support vector machines (SVM). Then we merged the SVM classifier (detailed in Supplementary material) with two classifiers [19,20], based on short-term features (calculated at the frame level), which characterize the spectral envelope and capture the articulatory disruptions. The fusion consisted on a majority vote of the three classes predictions, using the classifiers equal error rate thresholds as decision threshold.…”