Investigating Voice as a Biomarker for Leucine-Rich Repeat Kinase 2-Associated Parkinson's Disease

Voice as a Biomarker in Health-Tech: Mapping the Evolving Landscape of Voice Biomarkers in the Start-Up World

OBJECTIVE

The vocal biomarkers market was worth $1.9B in 2021 and is projected to exceed $5.1B by 2028, for a compound annual growth rate of 15.15%. The investment growth demonstrates a blossoming interest in voice and artificial intelligence (AI) as it relates to human health. The objective of this study was to map the current landscape of start-ups utilizing voice as a biomarker in health-tech.

DATA SOURCES

A comprehensive search for start-ups was conducted using Google, LinkedIn, Twitter, and Facebook. A review of the research was performed using company website, PubMed, and Google Scholar.

REVIEW METHODS

A 3-pronged approach was taken to thoroughly map the landscape. First, an internet search was conducted to identify current start-ups focusing on products relating to voice as a biomarker of health. Second, Crunchbase was utilized to collect financial and organizational information. Third, a review of the literature was conducted to analyze publications associated with the identified start-ups.

RESULTS

A total of 27 start-up start-ups with a focus in the utilization of AI for developing biomarkers of health from the human voice were identified. Twenty-four of these start-ups garnered $178,808,039 in investments. The 27 start-ups published 194 publications combined, 128 (66%) of which were peer reviewed.

CONCLUSION

There is growing enthusiasm surrounding voice as a biomarker in health-tech. Academic drive may complement commercialization to best achieve progress in this arena. More research is needed to accurately capture the entirety of the field, including larger industry players, academic institutions, and non-English content.

Current Practices in Voice Data Collection and Limitations to Voice AI Research: A National Survey

INTRODUCTION

Accuracy and validity of voice AI algorithms rely on substantial quality voice data. Although commensurable amounts of voice data are captured daily in voice centers across North America, there is no standardized protocol for acoustic data management, which limits the usability of these datasets for voice artificial intelligence (AI) research.

OBJECTIVE

The aim was to capture current practices of voice data collection, storage, analysis, and perceived limitations to collaborative voice research.

METHODS

A 30-question online survey was developed with expert guidance from the voicecollab.ai members, an international collaborative of voice AI researchers. The survey was disseminated via REDCap to an estimated 200 practitioners at North American voice centers. Survey questions assessed respondents' current practices in terms of acoustic data collection, storage, and retrieval as well as limitations to collaborative voice research.

RESULTS

Seventy-two respondents completed the survey of which 81.7% were laryngologists and 18.3% were speech language pathologists (SLPs). Eighteen percent of respondents reported seeing 40%-60% and 55% reported seeing >60 patients with voice disorders weekly (conservative estimate of over 4000 patients/week). Only 28% of respondents reported utilizing standardized protocols for collection and storage of acoustic data. Although, 87% of respondents conduct voice research, only 38% of respondents report doing so on a multi-institutional level. Perceived limitations to conducting collaborative voice research include lack of standardized methodology for collection (30%) and lack of human resources to prepare and label voice data adequately (55%).

CONCLUSION

To conduct large-scale multi-institutional voice research with AI, there is a pertinent need for standardization of acoustic data management, as well as an infrastructure for secure and efficient data sharing.

LEVEL OF EVIDENCE

5 Laryngoscope, 134:1333-1339, 2024.

Discovery and Analytical Validation of a Vocal Biomarker to Monitor Anosmia and Ageusia in Patients With COVID-19: Cross-sectional Study

Background

The COVID-19 disease has multiple symptoms, with anosmia and ageusia being the most prevalent, varying from 75% to 95% and from 50% to 80% of infected patients, respectively. An automatic assessment tool for these symptoms will help monitor the disease in a fast and noninvasive manner.

Objective

We hypothesized that people with COVID-19 experiencing anosmia and ageusia had different voice features than those without such symptoms. Our objective was to develop an artificial intelligence pipeline to identify and internally validate a vocal biomarker of these symptoms for remotely monitoring them.

Methods

This study used population-based data. Participants were assessed daily through a web-based questionnaire and asked to register 2 different types of voice recordings. They were adults (aged >18 years) who were confirmed by a polymerase chain reaction test to be positive for COVID-19 in Luxembourg and met the inclusion criteria. Statistical methods such as recursive feature elimination for dimensionality reduction, multiple statistical learning methods, and hypothesis tests were used throughout this study. The TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) Prediction Model Development checklist was used to structure the research.

Results

This study included 259 participants. Younger (aged <35 years) and female participants showed higher rates of ageusia and anosmia. Participants were aged 41 (SD 13) years on average, and the data set was balanced for sex (female: 134/259, 51.7%; male: 125/259, 48.3%). The analyzed symptom was present in 94 (36.3%) out of 259 participants and in 450 (27.5%) out of 1636 audio recordings. In all, 2 machine learning models were built, one for Android and one for iOS devices, and both had high accuracy—88% for Android and 85% for iOS. The final biomarker was then calculated using these models and internally validated.

Conclusions

This study demonstrates that people with COVID-19 who have anosmia and ageusia have different voice features from those without these symptoms. Upon further validation, these vocal biomarkers could be nested in digital devices to improve symptom assessment in clinical practice and enhance the telemonitoring of COVID-19–related symptoms.

Trial Registration

Clinicaltrials.gov NCT04380987; https://clinicaltrials.gov/ct2/show/NCT04380987

Assessing Parkinson's Disease at Scale Using Telephone-Recorded Speech: Insights from the Parkinson's Voice Initiative

Numerous studies have reported on the high accuracy of using voice tasks for the remote detection and monitoring of Parkinson’s Disease (PD). Most of these studies, however, report findings on a small number of voice recordings, often collected under acoustically controlled conditions, and therefore cannot scale at large without specialized equipment. In this study, we aimed to evaluate the potential of using voice as a population-based PD screening tool in resource-constrained settings. Using the standard telephone network, we processed 11,942 sustained vowel /a/ phonations from a US-English cohort comprising 1078 PD and 5453 control participants. We characterized each phonation using 304 dysphonia measures to quantify a range of vocal impairments. Given that this is a highly unbalanced problem, we used the following strategy: we selected a balanced subset (n = 3000 samples) for training and testing using 10-fold cross-validation (CV), and the remaining (unbalanced held-out dataset, n = 8942) samples for further model validation. Using robust feature selection methods we selected 27 dysphonia measures to present into a radial-basis-function support vector machine and demonstrated differentiation of PD participants from controls with 67.43% sensitivity and 67.25% specificity. These findings could help pave the way forward toward the development of an inexpensive, remote, and reliable diagnostic support tool for PD using voice as a digital biomarker.

Identifying individuals with recent COVID-19 through voice classification using deep learning

Recently deep learning has attained a breakthrough in model accuracy for the classification of images due mainly to convolutional neural networks. In the present study, we attempted to investigate the presence of subclinical voice feature alteration in COVID-19 patients after the recent resolution of disease using deep learning. The study was a prospective study of 76 post COVID-19 patients and 40 healthy individuals. The diagnoses of post COVID-19 patients were based on more than the eighth week after onset of symptoms. Voice samples of an ‘ah’ sound, coughing sound and a polysyllabic sentence were collected and preprocessed to log-mel spectrogram. Transfer learning using the VGG19 pre-trained convolutional neural network was performed with all voice samples. The performance of the model using the polysyllabic sentence yielded the highest classification performance of all models. The coughing sound produced the lowest classification performance while the ability of the monosyllabic ‘ah’ sound to predict the recent COVID-19 fell between the other two vocalizations. The model using the polysyllabic sentence achieved 85% accuracy, 89% sensitivity, and 77% specificity. In conclusion, deep learning is able to detect the subtle change in voice features of COVID-19 patients after recent resolution of the disease.

Detection of dementia on voice recordings using deep learning: a Framingham Heart Study

BACKGROUND

Identification of reliable, affordable, and easy-to-use strategies for detection of dementia is sorely needed. Digital technologies, such as individual voice recordings, offer an attractive modality to assess cognition but methods that could automatically analyze such data are not readily available.

METHODS AND FINDINGS

We used 1264 voice recordings of neuropsychological examinations administered to participants from the Framingham Heart Study (FHS), a community-based longitudinal observational study. The recordings were 73 min in duration, on average, and contained at least two speakers (participant and examiner). Of the total voice recordings, 483 were of participants with normal cognition (NC), 451 recordings were of participants with mild cognitive impairment (MCI), and 330 were of participants with dementia (DE). We developed two deep learning models (a two-level long short-term memory (LSTM) network and a convolutional neural network (CNN)), which used the audio recordings to classify if the recording included a participant with only NC or only DE and to differentiate between recordings corresponding to those that had DE from those who did not have DE (i.e., NDE (NC+MCI)). Based on 5-fold cross-validation, the LSTM model achieved a mean (±std) area under the receiver operating characteristic curve (AUC) of 0.740 ± 0.017, mean balanced accuracy of 0.647 ± 0.027, and mean weighted F1 score of 0.596 ± 0.047 in classifying cases with DE from those with NC. The CNN model achieved a mean AUC of 0.805 ± 0.027, mean balanced accuracy of 0.743 ± 0.015, and mean weighted F1 score of 0.742 ± 0.033 in classifying cases with DE from those with NC. For the task related to the classification of participants with DE from NDE, the LSTM model achieved a mean AUC of 0.734 ± 0.014, mean balanced accuracy of 0.675 ± 0.013, and mean weighted F1 score of 0.671 ± 0.015. The CNN model achieved a mean AUC of 0.746 ± 0.021, mean balanced accuracy of 0.652 ± 0.020, and mean weighted F1 score of 0.635 ± 0.031 in classifying cases with DE from those who were NDE.

CONCLUSION

This proof-of-concept study demonstrates that automated deep learning-driven processing of audio recordings of neuropsychological testing performed on individuals recruited within a community cohort setting can facilitate dementia screening.

Detection of dementia on voice recordings using deep learning: a Framingham Heart Study

BACKGROUND

Identification of reliable, affordable, and easy-to-use strategies for detection of dementia is sorely needed. Digital technologies, such as individual voice recordings, offer an attractive modality to assess cognition but methods that could automatically analyze such data are not readily available.

METHODS AND FINDINGS

We used 1264 voice recordings of neuropsychological examinations administered to participants from the Framingham Heart Study (FHS), a community-based longitudinal observational study. The recordings were 73 min in duration, on average, and contained at least two speakers (participant and examiner). Of the total voice recordings, 483 were of participants with normal cognition (NC), 451 recordings were of participants with mild cognitive impairment (MCI), and 330 were of participants with dementia (DE). We developed two deep learning models (a two-level long short-term memory (LSTM) network and a convolutional neural network (CNN)), which used the audio recordings to classify if the recording included a participant with only NC or only DE and to differentiate between recordings corresponding to those that had DE from those who did not have DE (i.e., NDE (NC+MCI)). Based on 5-fold cross-validation, the LSTM model achieved a mean (±std) area under the receiver operating characteristic curve (AUC) of 0.740 ± 0.017, mean balanced accuracy of 0.647 ± 0.027, and mean weighted F1 score of 0.596 ± 0.047 in classifying cases with DE from those with NC. The CNN model achieved a mean AUC of 0.805 ± 0.027, mean balanced accuracy of 0.743 ± 0.015, and mean weighted F1 score of 0.742 ± 0.033 in classifying cases with DE from those with NC. For the task related to the classification of participants with DE from NDE, the LSTM model achieved a mean AUC of 0.734 ± 0.014, mean balanced accuracy of 0.675 ± 0.013, and mean weighted F1 score of 0.671 ± 0.015. The CNN model achieved a mean AUC of 0.746 ± 0.021, mean balanced accuracy of 0.652 ± 0.020, and mean weighted F1 score of 0.635 ± 0.031 in classifying cases with DE from those who were NDE.

CONCLUSION

This proof-of-concept study demonstrates that automated deep learning-driven processing of audio recordings of neuropsychological testing performed on individuals recruited within a community cohort setting can facilitate dementia screening.

Remote Assessment of Parkinson's Disease Symptom Severity Using the Simulated Cellular Mobile Telephone Network

Telemonitoring of Parkinson's Disease (PD) has attracted considerable research interest because of its potential to make a lasting, positive impact on the life of patients and their carers. Purpose-built devices have been developed that record various signals which can be associated with average PD symptom severity, as quantified on standard clinical metrics such as the Unified Parkinson's Disease Rating Scale (UPDRS). Speech signals are particularly promising in this regard, because they can be easily recorded without the use of expensive, dedicated hardware. Previous studies have demonstrated replication of UPDRS to within less than 2 points of a clinical raters' assessment of symptom severity, using high-quality speech signals collected using dedicated telemonitoring hardware. Here, we investigate the potential of using the standard voice-over-GSM (2G) or UMTS (3G) cellular mobile telephone networks for PD telemonitoring, networks that, together, have greater than 5 billion subscribers worldwide. We test the robustness of this approach using a simulated noisy mobile communication network over which speech signals are transmitted, and approximately 6000 recordings from 42 PD subjects. We show that UPDRS can be estimated to within less than 3.5 points difference from the clinical raters' assessment, which is clinically useful given that the inter-rater variability for UPDRS can be as high as 4-5 UPDRS points. This provides compelling evidence that the existing voice telephone network has potential towards facilitating inexpensive, mass-scale PD symptom telemonitoring applications.

Voice for Health: The Use of Vocal Biomarkers from Research to Clinical Practice

Diseases can affect organs such as the heart, lungs, brain, muscles, or vocal folds, which can then alter an individual's voice. Therefore, voice analysis using artificial intelligence opens new opportunities for healthcare. From using vocal biomarkers for diagnosis, risk prediction, and remote monitoring of various clinical outcomes and symptoms, we offer in this review an overview of the various applications of voice for health-related purposes. We discuss the potential of this rapidly evolving environment from a research, patient, and clinical perspective. We also discuss the key challenges to overcome in the near future for a substantial and efficient use of voice in healthcare.

New insights into Parkinson's disease through statistical analysis of standard clinical scales quantifying symptom severity

Clinical research studies in Parkinson's Disease (PD) focusing on symptom assessment often rely on thorough time-consuming physical examinations quantified on clinical scales such as the Unified Parkinson's Disease Rating Scale (UPDRS). Although widely used in clinical research, realistic time constraints preclude its use in daily clinical practice. The Hoehn and Yahr (H&Y) staging is an alternative scale which is easier to administer and provides a succinct descriptor of overall PD severity. There is no universal agreement amongst neurologists on the specific PD symptoms they need to be assessing in order to prescribe treatments and optimize symptom management for their patients, and practically there are no clinical scales which are recorded in daily clinical practice. In this study, we systematically evaluate diverse symptoms (as expressed in 44 UPDRS items) and aim to provide a statistical association with UPDRS and H&Y using rank correlation and mutual information metrics. Moreover, we investigate the projection of a UPDRS item subset on a 2D plot to map onto H&Y. We report some statistically strong correlations of PD symptoms against UPDRS and H&Y (|R| ≥ 0.3), and provide an intuitively appealing visualization mapping onto H&Y. These findings may be useful to neurologists as practical guidance in their daily clinical routine.

LRRK2 in Parkinson disease: challenges of clinical trials

One of the most common monogenic forms of Parkinson disease (PD) is caused by mutations in the LRRK2 gene that encodes leucine-rich repeat kinase 2 (LRRK2). LRRK2 mutations, and particularly the most common mutation Gly2019Ser, are observed in patients with autosomal dominant PD and in those with apparent sporadic PD, who are clinically indistinguishable from those with idiopathic PD. The discoveries that pathogenic mutations in the LRRK2 gene increase LRRK2 kinase activity and that small-molecule LRRK2 kinase inhibitors can be neuroprotective in preclinical models of PD have placed LRRK2 at the centre of disease modification efforts in PD. Recent investigations also suggest that LRRK2 has a role in the pathogenesis of idiopathic PD and that LRRK2 therapies might, therefore, be beneficial in this common subtype of PD. In this Review, we describe the characteristics of LRRK2-associated PD that are most relevant to the development of LRRK2-targeted therapies and the design and implementation of clinical trials. We highlight strategies for correcting the effects of mutations in the LRRK2 gene, focusing on how to identify which patients are the optimal candidates and how to decide on the timing of such trials. In addition, we discuss challenges in implementing trials of disease-modifying treatment in people who carry LRRK2 mutations.

Developing a large scale population screening tool for the assessment of Parkinson's disease using telephone-quality voice

Recent studies have demonstrated that analysis of laboratory-quality voice recordings can be used to accurately differentiate people diagnosed with Parkinson's disease (PD) from healthy controls (HCs). These findings could help facilitate the development of remote screening and monitoring tools for PD. In this study, 2759 telephone-quality voice recordings from 1483 PD and 15 321 recordings from 8300 HC participants were analyzed. To account for variations in phonetic backgrounds, data were acquired from seven countries. A statistical framework for analyzing voice was developed, whereby 307 dysphonia measures that quantify different properties of voice impairment, such as breathiness, roughness, monopitch, hoarse voice quality, and exaggerated vocal tremor, were computed. Feature selection algorithms were used to identify robust parsimonious feature subsets, which were used in combination with a random forests (RFs) classifier to accurately distinguish PD from HC. The best tenfold cross-validation performance was obtained using Gram-Schmidt orthogonalization and RF, leading to mean sensitivity of 64.90% (standard deviation, SD, 2.90%) and mean specificity of 67.96% (SD 2.90%). This large scale study is a step forward toward assessing the development of a reliable, cost-effective, and practical clinical decision support tool for screening the population at large for PD using telephone-quality voice.