Your Voice Could Be the Next Coronavirus Test

Sars-Cov-2 tests are currently a hot topic, both for their ability to identify contagions and to assess if someone has already gone through the disease. In both instances, they are a critical tool to study the pandemic and to recover a resemblance of normality. The lack of available tests, however, and the high precision levels required from them, has dampened expectations regarding their large-scale use to fight the coronavirus. Nevertheless, there could be alternatives. The secret sauce would be artificial intelligence, which is already being used for the detection of the coronavirus on surfaces as well as in innovative technology projects aimed at analyzing contagion patterns. This time the approach is using AI to identify voice patterns in infected people. A Spanish company, in collaboration with the Basque Health Department, is already working on a solution that could leverage minor voice alterations that would go unnoticed for a human being.   

Despite the complexity of the technology, the process is comparatively simple. A while ago, we covered an innovative technology that detected heart attacks through smart speakers. The gasping and heavy breathing of someone affected by a heart attack were enough for artificial intelligence to identify a risky situation. Interestingly, the new coronavirus project was born when Spanis cardiologist Domingo A. Pascual noticed that people suffering from heart attacks experience changes in their voice tones. Now that phenomenon has been used to fight the coronavirus too.

Currently, the research team is recording voices and sound in hospital wards to create a sort of “library” that will include healthy, infected, and cured people. Once the sample has been validated, research will begin to establish the biometric markers among those three groups. The goal is to use the voice recognition system at doctor practices, but also through phone calls that could detect automatically infected people and refer them to their doctors. “We already know that COVID-19 disease affects the superior respiratory tract from a very early stage. The technology could thus prove especially relevant in the detection of asymptomatic cases, where quick and PCR tests are less efficient,” explained Pascual. Naturally, voice recognition would only be a preliminary test before carrying out more precise tests. However, it could substantially speed up detection processes.

Patterns of respiratory diseases

Artificial intelligence is uniquely able to detect patterns that appear in many illnesses and diseases. Currently, several studies have collected markers from around twenty respiratory diseases. Therefore,  the technology’s potential is not limited to coronavirus and can have applications for other pathologies.   

The goal for the researchers of this artificial intelligence project is to make their findings available to the scientific community at large, as well as public institutions. The developers expect to have the first detection percentages ready in a matter of three or four months.

Source: El Confidencial