JMIR Publications

ABSTRACT

Background:

The Covid-19 disease has multiple symptoms, being anosmia, varying from 75-95%, and ageusia, varying from 50-80% of infected patients, the most prevalent ones. An automatic assessment tool for these symptoms will help monitoring the disease in a fast and non-invasive 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 is made on population-based data. Participants were assessed daily through an online questionnaire and asked to register two different types of voice recordings, they were adults (older than 18 years old) that were confirmed by a PCR test to be positive to Covid-19 in Luxembourg and that passed through the exclusion criteria. Statistical methods like dimensionality reduction, multiple statistical learning methods and hypothesis tests were used throughout this study.

Results:

This study included 259 participants. Young (<35 years old) and females showed a higher rate of ageusia and anosmia. Participants were 41 (SD = 13) years old on average and the dataset was balanced for sex (54% females and 48% males). The analyzed symptom was present in 36% of the population and in 28% of audio recordings. Two machine learning models were built, one for Android and one for iOS devices and both had high accuracy, being 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 it. Upon further validation, these vocal biomarkers could be nested in digital devices to improve symptom assessment in clinical practice and enhance telemonitoring of Covid-19-related symptoms.