Q&A: Vocal biomarkers may be used to monitor patients with COVID-19

With a simple 25-second voice recording, researchers may be able to monitor symptom resolution in patients with COVID-19.

Guy Fagherazzi, PhD, ADR, the director of the department of precision health and group leader of the deep digital phenotyping research unit at the Luxembourg Institute of Health, and colleagues found that they could use a machine learning model to derive a vocal biomarker. They could then use that information to “monitor the resolution of COVID-19-related symptoms with elevated accuracy and calibration,” the researchers wrote in their study, published in PLOS Digital Health.

As a “non-invasive, quick, and cheap way” to remotely monitor patients with COVID-19, “such a vocal biomarker could be integrated into future telemonitoring solutions, digital devices, or in clinical practice for a rapid screening during a consultation to aid clinicians during anamnesis,” they wrote.

Fagherazzi spoke with Healio about the mechanisms behind the technology and its potential clinical implications.

Healio: Can you explain how vocal biomarkers can be used to assess someone’s COVID-19 status?

Fagherazzi: People experience changes in their voice production when a symptom or a disease is present. Sometimes, these changes are not really perceptible with the human ear, but with the recent advancements in audio signal processing, digital technologies and artificial intelligence, we are now able to predict some health issues based on voice recordings. For COVID-19, frequent symptoms, such as fatigue, sore throat, respiratory issues, fever, etc., induced some changes in the voice that we are able to capture to predict the symptomatic status of a given individual.

Healio: Can these biomarkers reveal whether a patient has long COVID?

Fagherazzi: For now, our research demonstrated that it is feasible to detect symptoms in people with COVID-19. Some, but not all, people with long COVID may experience the same symptoms that persist over time for several months, so we hope that our technology will also enable these people to monitor the evolution of long COVID symptoms over time.

Healio: Can these biomarkers be an indicator for the severity of a patient’s illness, or their symptoms?

Fagherazzi: This work aims at predicting the symptomatic status and not the severity of the disease.

Healio: How effective is the technology? How does this compare with other COVID-19 monitoring tools that are used in practice?

Fagherazzi: The technology allows a detection of COVID-19-related symptoms with good performances. The aim is not to diagnose COVID-19 (so we do not try to compete with PCR testing, for instance) but rather to use it 1) as a triage or screening solution, and 2) as a monitoring solution to facilitate remote patient monitoring.

Healio: Will there ever come a day where providers can monitor their patients’ COVID-19 status in practice using this technology?

Fagherazzi: Hopefully, we will soon be in the position to validate our technology in other contexts and accelerate the development of this technology to bring it to the market in the coming years.

Healio: Other companies are also investigating the use of vocal biomarkers in diagnosing COVID-19. How similar is this to your technology? How does it differ?

Fagherazzi: The approach is similar as they are also using voice recorded on study participants, but the ultimate objective is different. They want to diagnose COVID-19. On our end, we are working only on people with PCR-positive COVID-19 and try to predict whether these people have symptoms or not.

Healio: What other health conditions could potentially be assessed by vocal biomarkers?

Fagherazzi: Many other conditions or symptoms will soon be detected and monitored throughout voice. In our lab, we are working on vocal biomarkers to detect depressive symptoms, quantify respiratory quality of life, fatigue, flare-ups of immune related diseases and diabetes.

Healio: Is there anything else you’d like to add?

Fagherazzi: We are still looking for research volunteers to improve this technology. People can still contribute to this exciting research topic by donating their voice on the Colive Voice platform. It is anonymous, it takes 20 minutes, and it will help improve the monitoring of health.

References:

• Fagherazzi G, et al. PLOS Digit Health. 2022;doi:10.1371/journal.pdig.0000112.
• The sound of COVID-19. https://www.lih.lu/wp-content/uploads/2022/10/PR_The-Sound-of-COVID-19_EN.pdf. Published Oct. 24, 2022. Accessed Oct. 27, 2022.