Development of Voice Biomarkers of Frequent COVID-19 and Long Covid-related Symptoms Based on Data From Users of the Long COVID Companion App

Five years after the start of the COVID-19 pandemic, it has been estimated that a mean of 10-20% of COVID-19 patients will develop Long COVID, which represents at least 65 million people worldwide. In Luxembourg, based on the current status of the COVID-19 pandemic and the prevalence of Long COVID in other countries, we estimate that approximately 25,000 individuals may have been affected by Long COVID.

People with Long COVID (PWLC) present complaints such as tachycardia, extreme fatigue, dyspnea, and inability to perform daily physical tasks. More than 200 symptoms have been associated with Long COVID, and multiple organs are affected. Our previous work showed that 59% of COVID-19-infected people from the PrediCOVID cohort study reported 1 or more persisting symptom(s) after one year. The number of persisting symptoms increased with the initial disease severity, and the quality of life of those participants was notably impacted by sleep disorders (54%) and compromised respiratory function (12.9%). Nonetheless, individuals with an initially asymptomatic or mild form of COVID-19 infection could also be affected.

Long COVID care, as for other chronic diseases, should align with the concept of minimally disruptive medicine, aiming for a reduced burden on patients' lives while maximizing health outcomes. PWLC frequently have several healthcare professionals in charge of different aspects of their care, with many appointments and travels to manage. They are also regularly asked to complete long standardized questionnaires or scales to evaluate their symptoms, which generates an avoidable burden on their lives if care is not coordinated. The development of innovative methods to integrate multiple Patient Report Outcomes in a portable, versatile way, to reduce travels for medical care, and overall for remote health's monitoring is therefore of the highest importance.

Few existing apps that could meet the needs of PWLC already exist, such as "Visible" or "Living With". However, these apps are only available in the US and in the UK, respectively. Furthermore, the "Living with" app is available only by invitation. This limits their availability for PWLC in Europe. Some other apps designed for other chronic conditions, like myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) Pacing app, are also used by some PWLC to manage their daily energy levels. However, these apps do not fully meet the specific needs of PWLC, and are only available in specific countries, in English, and some of them only on the iOS operating system, which restricts their potential use.

Voice is a promising candidate increasingly used in mobile health (mHealth) interactions in chronic diseases. Voice is an easy and cheap medium to collect and can be easily integrated into a device like a smartphone, now widely used by people of different ages and education levels. It can be used for vocal biomarker assessment as new clinical endpoints for relevant symptoms. Vocal biomarker candidates have already been identified by the Deep Digital Phenotyping (DDP) research team, based on data from the PrediCOVID cohort study, with performances above 80% to detect fatigue, loss of taste and smell, and symptomatic status in COVID-19-infected people. However, further progress is required before these vocal biomarkers can be used in clinical practice. New vocal biomarkers of global health status or of other symptoms need to be developed to allow a more personalized monitoring of PWLC.

The LIH developed the Long COVID Companion app based on the results of UpcomingVoice co-design study and in collaboration with the ApresJ20 Long COVID patient association in France. The app is the result of a participative process involving patients and healthcare professionals in charge of PWLC patients. App users can monitor their health and symptoms on a daily basis or at the frequency they want. App also offers the possibility to complete a daily life and a medical diary, a voice journaling, and to generate PDF reports with graphical visualization of symptom evolution. A new module dedicated to research with standardized voice recordings has been integrated in the app for the users who will take part in the present study The app was released in April 2024 and has more than 1700 users as of the end of March 2025.

It is of highest importance to assess the evolution of health of PWLC for a better understanding of the disease. Results will allow the development of personalized monitoring strategies by identifying vocal biomarkers of Long COVID symptoms and of the global health of PWLC.

As fatigue is the most frequent and impairing symptom of PWLC it is crucial to better understand the drivers of fatigue and to develop monitoring strategies.

The objectives of this study are :

Primary objective: To develop vocal biomarker candidates for the main Long COVID symptoms (fatigue, brain fog, respiratory problems, sleep issues, stress, anxiety,..) in a population of people with Long COVID.

Secondary objectives:

• to assess the intra-individual longitudinal evolution of voice characteristics of people with LC
• to assess app usability and acceptability in the long-term.