Can be detected COVID-19 from the voice of patients? The company Vocalis claims that thanks to AI, this is possible.
Detection of COVID-19 without testing is not an easy task and this is mainly due to the fact that the symptoms vary greatly from person to person. If someone has symptoms such as sore throat, fever, cough or loss of smell or taste, they should look for it.
Many companies try to facilitate this process by utilizing simple technologies such as mobile devices apps, which claim to detect COVID-19. One of these companies is Vocalis based in Israel.
What do these technologies help with?
Diagnosis for COVID-19 or any other disease is difficult unless you go to a testing center for testing. To make it easier for people, Vocalis has created an app based on artificial intelligence (AI) and examines the vocal characteristics of individuals to determine if they are suffering from a disease without having to visit one hospital.
This is an interesting approach, given that other methods, such as thermometry, can not detect asymptomatic patients. Vocalis solution could help in these cases.
How does AI help?
Vocalis collects a lot of data related to a person's voice. He has developed an AI technology that determines if a person has a disease, based on the analysis of his voice.
The company algorithm extracts 512 features from a voice sample and creates an image - a spectrogram - with which it can diagnose possible diseases.
When a company wants to detect a specific disease, it first conducts clinical trials to identify what kind of recordings it will take to identify it and what sound elements it will need to detect.
For example, to detect COVID-19, the AI must identify voice features that could be associated with symptoms such as fever, headache, and shortness of breath.
In February, Vocalis participated in a joint study with Municipal Corporation of Greater Mumbai in India, which examined 2.000 participants.
They spoke several languages, such as English, Hindi, Marathi and Gujarati. The company installed the tools VocalisCheck on the smartphones of the participants to measure the effectiveness of algorithms. In the end, the researchers found that the algorithm was 81,2% accurate in detecting COVID-19. Many of the patients were asymptomatic.
Vocalis tools are currently being developed at USA, The Chile, The Ινδονησία, The South Africa, The Romania and Luxembourg.
The company wants to get more datasets to make COVID-19 detection even more accurate. In addition, it wants to offer its tools in more areas.