Sber's artificial intelligence laboratory has created an algorithm that can recognize
As Sber explained, the records are transformed intoa spectrogram that shows the energy of sound at different frequencies. This data is analyzed by a deep convolutional neural network, which was trained on open data. More than a thousand samples of breathing and cough recordings of Russian patients were used as the latter.
Sberbank clarified that the system does not work with suchthe same accuracy as PCR, “but already has comparable characteristics.” The program acts rather as a personal daily test for coronavirus. However, it is not a medical instrument.
In the near future, Sberbank will create a corresponding application. It will appear in the App Store and Google Play.