In terms of epidemiology, people infected with COVID-19 are divided into three groups:
1
Those infected with symptoms - are at home or in hospital, making a minimal contribution to the spread of the epidemic;
2
COVID-infected dissidents - do not self-isolate, with mild symptoms, spread the virus, but this group is small;
3
Asymptomatic carriers - make a major contribution to the spread of the epidemic.
the artificial intelligence model could distinguish between coughs of asymptomatic COVID positive carriers and coughs of healthy people
At the end of September 2020, a study was published by a group of scientists at MIT University in Boston, stating that the artificial intelligence model they had created could distinguish between coughs of asymptomatic COVID positive carriers induced at the request of researchers and coughs of healthy people.
The model is 98.5% accurate, which exceeds the accuracy of laboratory tests. PCR smear accuracy is 70-80% with preanalytical features taken into account.
The Russian Institute of Physics of the Lebedev Russian Academy of Sciences also has a developed model, its accuracy is 89% today.
At the end of September 2020, a study was published by a group of scientists at MIT University in Boston, stating that the artificial intelligence model they had created could distinguish between coughs of asymptomatic COVID positive carriers induced at the request of researchers and coughs of healthy people.
The model is 98.5% accurate, which exceeds the accuracy of laboratory tests. PCR smear accuracy is 70-80% with preanalytical features taken into account.
The Russian Institute of Physics of the Lebedev Russian Academy of Sciences also has a developed model, its accuracy is 89% today.
This means that there is an instant, non-invasive, accurate and free COVID test.
In Russia, 20-25 thousand people a day get sick. The virus is released in the external environment according to different data from 8 to 11 days. Thus, at the moment 300-400 thousand people are contagious. Adding asymptomatic and uncounted - will be about a million.
If we all cough, for example, in an application on your cell phone, and put to quarantine only the distributors for 10 days - the epidemic COVID-19 can be taken under control or significantly slowed down in a few weeks.
We conducted scientific research and collected a dataset of induced cough records of people with COVID-19, confirmed by a test, of asymptomatic carriers, as well as a control group of records of healthy people. It has been made publicly available.

It is the only open public data set in the world today with such quantity and quality of records.
In this regard, we call for:
The developers of neural networks:
A dataset is enough for model development and training. It can be found HERE. It is necessary to make the model as soon as possible and make the code available to the public so that any developer of mobile applications can use it for integration into their solutions. Send us the results, we will place the models in the open access in the same repository.
Large IT companies, whose applications are installed on the phones of most citizens:
The technology exists in Russia. The Lebedev Institute of Physics of the Russian Academy of Sciences can share it with those who want it. Implementation of the technology in Yandex Alice, Sberbank and Vkontakte applications will provide quick access to it for almost the entire population.
Governments:
Introduce mandatory selective (not mass!) 10-day self-isolation only of those citizens who are carriers, with administrative responsibility for its violation. Verification can be carried out by law enforcement officials using cell phones.
Every day costs thousands of lives!
Get in touch with us: