Mobile App Can Detect COVID-19 Infection in People Through Their Voices Using AI

A new study recently showed how artificial intelligence could be used to identify COVID-19 infection in the voices of people by means of a mobile app.

As specified in a EurekAlert! report, the new research, to be presented today at the European Respiratory Society International Congress in Barcelona, Spain; the AI model the researchers used, is more precise compared to the lateral flow or rapid antigen tests and is more affordable, fast, and easy to use.

This means that it can be used in low-income nations where PCR tests are costly or difficult to distribute.

Researcher Wafaa Aljbawi, from the Institute of Data Science, Maastricht University, The Netherlands, said during the congress the AI model was precise 89 percent of the time, while the lateral flow tests' accuracy differed according to the brand.

More so, lateral flow tests were substantially less precise in detecting COVID-19 infection in those who did not exhibit any symptoms.

AI Algorithms Used to Detect COVID-19 Infection

Such promising results suggest that simple voice recordings are "fine-tuned AI algorithms that can potentially attain high precision" in identifying which patients are infected with COVID-19, explained the researcher. She added, that such tests can be provided at no cost, not to mention simple to interpret.

Furthermore, they allow remote, virtual testing and have a turnaround time of shorter than one minute. They could be utilized, for instance, at the entry points of large gatherings, allowing for fast screening of the population.

Typically, COVID-19 impacts the upper respiratory tract and the vocal cords, resulting in a person's voice.

Aljbawi, with her supervisors, Dr. Sami Simons, a pulmonologist at Maastricht University Medical Centre, and Dr. Visara Urovi, also from the Institute of Data Science, decided to examine if it was plausible to use AI to analyze voices to detect COVID-19.

They utilized data from the University of Cambridge's crowd-sourcing COVID-19 Sounds App, containing over 890 audio samples from over 4,500 healthy and non-healthy participants, of whom 308 had tested positive for COVID-19.

Voice Analysis Used

Essentially, the app is installed on the user's mobile phone; the participants report some basic information about medical history, demographics, and smoking status, and then are asked to record some respiratory sounds.

These include coughing thrice, breathing deeply through the mouth about three to five times, and reading a short sentence on the screen thrice.

Moreover, in their study published in The Lancet Respiratory Medicine, the research team used a voice analysis method known as "Mel-Spectrogram analysis," which identifies different features of voice like power, variation, and loudness over time.

In this manner, explained Aljbawi, they can decompose the numerous properties of the participants' voices.

In order to distinguish the COVID-19 patients' voices from those who did not have the disease, the researchers developed different AI models and evaluated which one worked best at categorizing the COVID-19 cases.

MyCOPD Mobile App

Called myCOPD, the mobile app is a cloud-based interactive app developed by clinicians and patients and is available for use in the National Health Service of the United Kingdom.

It was established in 2016 and, thus far, has more than 15,000 COPD patients using it to help them manage their infection.

The study investigators collected more than 45,600 records for over 180 patients from August 2017 to December 2021. Out of this total, 45,007 were records of stable illnesses, and 629 were worsenings.

Exacerbation predictions were produced one to eight days before a self-reported exacerbation. The team used the data to train models on 70 percent of the data and have it tested on 30 percent.

According to a similar Bioengineer.org report, the patients were "high engagers" who had been using the app every week for months or even years to record their symptoms and other health information, set reminders, record medication, and a clinician dashboard, allowing them to offer insight, co-management, as well as remote monitoring.

Information about the myCOPD app is shown on Townhill Community Surgery's YouTube video below:

Check out more news and information on COVID-19 and Technology in Science.