A recent study to be presented this Monday at the International Congress of the European Respiratory Society in Barcelona, Spain, determined that COVID-19 could be detected through a new artificial intelligence (AI) model. through someone's voice through a mobile phone application.
The AI model used in this study is cheaper, faster, and easier to use than rapid antigen/lateral flow tests, according to the News Medical report .
This would be a great advantage suitable for application in low-income countries where PCR tests are expensive or difficult to perform.
The accuracy of lateral flow tests varies widely by brand, but the AI model was accurate 89% of the time, according to Wafaa Aljbawi, a researcher at the Institute for Data Science at Maastricht University in the Netherlands.
It added that lateral flow tests had a significantly lower ability to identify COVID-19 infection in people without symptoms.
The diagnosis of COVID with just a voice recording
These encouraging results indicate the potential for basic voice recordings and custom AI algorithms to achieve high accuracy in identifying patients with COVID-19 infection. Such tests can be provided free of charge and are easy to interpret, according to Aljbawi.
The upper respiratory tract and vocal cords are usually affected by COVID-19, altering a person's voice. Dr. Visara Urovi, also from the Institute of Data Science, and Dr. Sami Simons, a pulmonologist at Maastricht University Medical Center tested the feasibility of using AI to analyze voices to identify COVID-19.
The team used information from Cambridge University's COVID-19 Sounds app, which includes 893 audio samples from 4,352 healthy and unhealthy subjects, of whom 308 received positive COVID-19 test results.
Users must download the app on their smartphones, provide basic demographic, medical, and smoking status information. They then record their breath sounds.
Coughing three times, taking deep breaths through the mouth three to five times, and reading a short sentence on the screen three times are some of the steps users must take to get tested.
The researchers used a method of analyzing speech, known as Mel's spectrogram analysis, which distinguishes various characteristics of speech such as loudness, power, and fluctuation over time.