Scientists from the Institute of Data Science at Maastricht University in the Netherlands have started a study on whether Kovid-19 can be detected in human voices with a mobile application using artificial intelligence.
In the study, 893 audio recordings (308 of which were positive for Kovid-19) were analyzed from 4,353 healthy and unhealthy participants in Cambridge University’s Kovid-19 Audio Application.
To analyze the recordings, the experts used a scale called the Mel-spectrogram, which shows the change in sound frequencies of the human ear.
Wafaa Aljbawi, from the Institute of Data Science at Maastricht University, said, “We have developed different artificial intelligence models to distinguish the voices of those who do not have COVID-19 from other voices, and we evaluated which model gives better results in detecting Kovid-19 cases.” said.
Aljbawi noted that the Long Short-Term Memory (LSTM) model performed better than other artificial intelligence models.
In the research, it was noted that it provides 89 percent accuracy in detecting Kovid-19 cases with sound, and the accuracy of detecting people who do not have Kovid-19 patients is 83 percent.
The results of the study were presented at the International Congress of the European Respiratory Society held in Barcelona, Spain.
Researchers stated that this application is faster, easier and cheaper than antigen tests in the diagnosis of Kovid-19. (AA)