The Netherlands’ Maastricht University has made a smartphone app that can recognize the sound of COVID-19 in a person’s voice. The developers say that the app’s ability to find problems is “more accurate than lateral flow tests.”
The COVID-19 Voice App is free to download from the App Store and also Google Play.
A common sign of a COVID-19 infection is inflammation of the respiratory system and vocal cords. Which can make the patient’s voice change. The researcher is trying to find out if these symptoms serve to make the right diagnosis.
With an accuracy rate of more than 89%, the artificial intelligence (AI) that the research team used is more reliable than quick antigen tests. The AI model did teach with more than 893 audio samples from 4,352 healthy and sick people. 38 of them tested positive for COVID-19.
Dr. Wafaa Aljbawi, a researcher at Maastricht University’s Institute of Data Science, said:
These encouraging results show that simple voice recordings and well-tuned AI algorithms are sometimes like to discover patients with Covid-19 infection with high accuracy. These tests are easy to understand and don’t cost anything. They also permit digital testing to take place from a distance and offer a return time of less than a minute.
Monday, the results are reported at the International Congress of the European Respiratory Society in Barcelona.
How the application works
After collecting basic information like demographics, medical history, and smoking status. The program asks the user to record a sample of breathing sounds. This includes things like reading a short sentence three times on the screen, reading a short statement aloud three times, and coughing three times.
Mel-spectrogram analysis is one way that the app works. It looks at vocal characteristics like volume, intensity, and changes in timing to make a diagnosis.
As previously said, the software is 89% accurate at finding positive situations and only 83% accurate at finding negative ones.
When compared to other, more advanced diagnostic methods like the lateral flow test. These results show that Covid-19 can be diagnosed with much more accuracy. To put it another way, out of every 100 people tested, the lateral flow test would miss 44, while the LSTM model, which uses artificial intelligence, would only miss 11.
Read more: Telecom companies’ high costs are a major hurdle to 5G expansion in Pakistan