The AI-powered technology is easier to use and more accurate than a lateral flow test, scientists say. The mobile app takes less than a minute to flag positive cases and gives an accurate result 89 percent of the time and negative cases 83 percent. The accuracy of lateral flow tests varies widely and can miss between 20 percent and 81 percent of positive cases in different settings, according to Imperial College London. The new app could be used to very quickly check people for the bug before they attend mass events such as concerts and major sporting events. AI-powered technology is easier to use and more accurate than a lateral flow test, scientists say It could also be developed in poorer countries where gold PCR tests are very expensive and often difficult to distribute. Dutch researchers say the coronavirus usually affects the upper respiratory tract and vocal cords, leading to changes in a person’s voice. The team decided to investigate whether it was possible to detect the new virus in people’s voices. The experts used data from Cambridge University’s Crowdsourcing COVID-19 Sounds app, which contained 893 sound samples from 4,352 participants. And from the sample study, 308 of the respondents tested positive for the coronavirus.
The app is installed on the user’s mobile phone and participants report some basic details about demographics, medical history and smoking status. They are then asked to record some breathing sounds that include coughing three times, breathing deeply through the mouth three to five times, and reading a short sentence on the screen three times. The researchers used a voice analysis technique called Mel-spectrogram analysis, which identifies different voice characteristics such as pitch, power and variation over time. To distinguish the voices of patients with COVID-19 from those who did not have the disease, the team built different artificial intelligence models and evaluated which one worked best in classifying positive cases. A model called Long-Short Term Memory (LSTM) outperformed the others. Since the start of the project, 53,449 audio samples from 36,116 participants have been collected and can be used to improve and validate the model’s accuracy It is based on neural networks, which mimic the way the human brain works and recognize the underlying relationships in the data. It works with sequences, which makes it suitable for modeling signals collected over time, such as from voice, due to its ability to store data in its memory. Researcher Wafaa Aljbawi, from Maastricht University, said: “These promising results suggest that simple voice recordings and accurate artificial intelligence algorithms can potentially achieve high accuracy in identifying patients with COVID-19 infection. “Such tests can be provided at no cost and are easy to interpret. “Plus, they enable remote, virtual testing and have a turnaround time of less than a minute. “They could be used, for example, at entry points for large gatherings, allowing rapid population control. “These results show a significant improvement in the accuracy of diagnosing COVID-19 compared to state-of-the-art tests such as the lateral flow test. “The lateral flow test has a sensitivity of only 56 percent, but a higher specificity rate of 99.5 percent. “This is important as it means that the lateral flow test misclassifies infected people as negative for COVID-19 more often than our test. “In other words, with the AI ​​LSTM model, we could miss 11 out of 100 cases that would continue to spread the infection, while the lateral flow test would miss 44 out of 100 cases. “The high specificity of the lateral flow test means that only one in 100 people would be falsely said to be positive for COVID-19 when, in fact, they were not infected, whereas the LSTM test would misdiagnose 17 in 100 uninfected people as positive. “However, since this test is almost free, it is possible to invite people for PCR tests if the LSTM tests show that they are positive.” The team says further research needs to be done with more participants before the app starts appearing on users’ phones. Since the start of the project, 53,449 audio samples from 36,116 participants have been collected and can be used to improve and validate the accuracy of the model. The team is also conducting more analysis to understand which parameters in the voice affect the AI ​​model. The findings will be presented at the European Respiratory Society International Congress in Barcelona, ​​Spain.