According to researchers, a mobile phone app can detect COVID in people’s voices with “high accuracy” using artificial intelligence (AI).
An AI model is said to be 89% accurate and cheap to use, meaning it could be applied in low-income countries where PCR tests are more expensive.
The scientists say results can be provided in less than a minute and are said to be a “significant improvement” in the accuracy of lateral flow tests.
The infection usually affects the upper respiratory tract and vocal cords, so the researchers decided to analyze changes in the voice using an AI model to detect it. COVID.
“These promising results show that simple voice recordings and fine-tuned AI algorithms have a potential,” said Wafaa Aljbawi, a researcher at the Institute of Data Science at Maastricht University in the Netherlands. can achieve high accuracy in identifying which patients are infected with COVID-19.
“Such tests can be freely provided and simple to interpret. Furthermore, they allow for remote, virtual testing and have a turnaround time of less than a minute.
“They could, for example, be used at entry points for mass gatherings, allowing for rapid population screening.”
Data used from the University of Cambridge’s community sourcing COVID19 Sounds app. This included 893 audio samples from 4,352 healthy and unhealthy people.
Users need to provide information about their medical history, smoking status and demographics and record some respiratory sounds, such as coughing and reading a short sentence.
A speech analysis technique – called Mel-spectrogram – identified different speech characteristics to “decompose many of the characteristics of the participants’ voices”.
Ms Aljbawi added: “These results show a significant improvement in the accuracy of COVID-19 diagnosis compared with modern tests such as lateral flow tests.
Variant piercing program starts in UK
“The side-flow test has only 56% sensitivity, but the specificity rate is higher than 99.5%. This is important because it shows that the side-line test is misclassifying infected people as COVID- 19 negative more often than our test.
“In other words, with the AI LSTM model, we might miss 11 out of 100 cases that would go on to infect, while lateral flow testing would miss 44 out of 100 cases.”
The AI model is also being used for an application to predict exacerbations of chronic obstructive pulmonary disease.
The study will be presented to the European Respiratory Society International Congress in Barcelona on Monday.