The structure of Rosatom will create a model for the spread of epidemics

By the end of the year, the structure of Rosatom will create a Data Farm - software that will predict the spread of infections based on data from Sberbank and MTS. An accurate forecast requires data from other sources,

"Daughter" of the state corporation "Rosatom" - Federal State Unitary Enterprise "Russian Federal Nuclear Center - All-Russian Research Institute of Technical Physics. Academician E.I. Zababakhin (VNIITF) — creates a Data Farm to automatically predict the spread of viral infections using artificial intelligence technology and based on data from MTS and Sberbank, as well as from the Internet and other sources. This follows from the FSUE tender materials posted on the public procurement website.

This is the third such purchase, the previous two were carried out in 2020 and 2021. As stated in the terms of reference, “in the context of the emergence of epidemics, an adequate medium- and long-term description of their distribution is necessary” across RUSSIA, which takes into account the introduction and lifting of quarantine, the psychological and behavioral characteristics of various population groups, the impact on them of the slowdown and gradual restoration of work activity, and also vaccination. To solve this problem, it is necessary to create a "mathematical model that is closest to the real situation."

Up-to-date data on CORONAVIRUS in Russia and the world. Infographic Society

At the current stage of the project, it is necessary to conduct trial operation of an experimental sample of the Data Farm software package: to work out technologies for extracting, processing and storing initial data using artificial intelligence, and also to create a database to determine the model of the spread of viral infections in cities from the list agreed with the customer. What cities will be discussed, as well as under what conditions and what kind of data MTS and Sber will transmit, as well as from what other sources the program will receive information, are not specified in the procurement materials.

A representative of VNIITF told RBC that the purchase is being carried out to carry out work under a grant that was issued back in 2020 to simulate epidemics of viral infections. We are talking about the amount of 300 million rubles, which was received by the consortium, which, in addition to VNIITF, included the Institute of Applied Physics, the Institute of Applied Mathematics. M.V. Keldysh, National Research Center for Epidemiology and Microbiology named after N.N. N.F. Gamalei, National Medical Research Center for Phthisiopulmonology and Infectious Diseases. The project is led by Corresponding Member of the Russian Academy of Sciences, DOCTOR of Physical and Mathematical Sciences Sergey Lebedev. The VNIITF representative did not answer other questions from RBC.

Bill Gates assessed the danger of monkeypox and called the timing of the new pandemic Society

How Rosatom predicted the spread of covid

As VNIITF reported in May 2020, the most popular SEIRD model in the world was upgraded to predict the spread of coronavirus, from which it follows that if the rate of infection is higher than the rate of recovery, then the epidemic grows, and if it is lower, then it fades.

To extinguish the epidemic, it was necessary to reduce the rate of infection by at least four times. Initially, VNIITF scientists modeled various options for reducing the speed by increasing the level of immunity of the population, introducing quarantine, etc. But since a significant part of the population could develop COVID asymptomatically, other parameters had to be taken into account: which group a person belongs to (children, students, working, pensioners), if he works, then what is the size of the team, etc.

Such a model would allow, for example, to analyze the introduction of quarantine in relation to certain social groups or in certain areas of activity (work, transport, shops). For such a model to work, adequate initial data were required - from the population and its distribution by social groups to the workload of various modes of transport or shops. The VNIITF report indicated that Sberbank and MTS offered assistance in obtaining operational statistics on certain aspects of people's behavior and its changes during quarantine. “So, Sberbank, based on the analysis of the flow of transactions on plastic cards, can assess the change in the activity of people in stores, and MTS, using the geolocation capabilities of cellular communications, can provide generalized information on movements,” the message said.

The government became interested in the work, and a grant was allocated for this study.

Read on RBC Pro Pro How to relax,to have time to restore strength over the weekend Instructions Pro How to return money or goods,who are stuck in Europe Instructions Pro Consumers cut costs.In what niches does it make sense to open a business?Will NSD succeed in unfreezing foreign shares of Russians in the EU court?Pro It's time for factories to forget about development and learn to copy foreign products Articles Pro Q-commerce:what awaits fast delivery services Articles

Why you need a Data Farm

It follows from the procurement materials that the contractor will have to prepare an experimental sample of the Data Farm software package for trial operation (including the development of options for obtaining data), conduct a pilot, and then evaluate its results. The initial price of the contract is 17 million rubles. Earlier, in 2020, by order of VNIITF, the Data Farm concept was created, and in 2021, a technical solution based on open source software, software modules for the future complex, etc. were developed. In previous years, VNIITF chose as the only contractor in tenders Institute of Applied Mathematics. M.V. Keldysh.

According to an interlocutor of RBC, close to one of the project participants, when building its model, VNIITF can use impersonal tracks of the trajectory of people's movement. “The only way to model the spread of infections in this case is to calculate the intersection of infected and uninfected citizens,” he said.

Virologists assessed the possibility and timing of a new wave of covid-19 Society

Another expert familiar with such studies noted that such predictive models use data that directly accompany or influence the development of the phenomena they estimate: in the case of an infection, this is the number of inhabitants, the density, number and intensity of social contacts, traffic in the city and country as a whole, retrodata on the number of cases, recovered, vaccinated, etc. “The larger the slice of data, the more accurate the forecast. All models in one way or another differ in the types and amount of data that are needed to build a forecast, the speed of their processing, and, most importantly, the accuracy of predicting the result. If the model is accurate, there are chances that it will be used when making decisions on infection control measures, when introducing quarantine, planning the procurement and transportation of medicines, and calculating the burden on the healthcare system, ”

A representative of Sberbank told RBC that their research team “carried out joint research in the field of predicting the epidemiological dynamics of COVID, including the exchange and comparison of the results of various simulation models.” “The exchange of personal data, information constituting commercial and banking secrets, and any other confidential information has never been carried out,” he insists.

A spokesman for MTS declined to comment. RBC sent a request to the Institute of Applied Mathematics. M.V. Keldysh and the Ministry of HEALTH.

How accurate will the forecast be?

Virologist, Corresponding Member of the Russian Academy of Sciences Alexander Lukashev noted that in Russia a number of scientific institutions have been developing models for predicting the spread of infections in recent years, since they can be in high demand. “The reason is that we do not have up-to-date infection prediction models. Those that were developed in the twentieth century turned out to be ineffective, because they grew out of physics and chemistry, where many parameters of molecules or atoms are identical, while all people are different in characteristics, ”the expert says.

According to Lukashev, such development should be done by multidisciplinary teams - not only mathematicians, but also doctors with sociologists. However, only the data that Sberbank and MTS can provide will not be enough for such work. “It is also important to take into account the different contagiousness of patients, the different social activity of people, their different susceptibility to the disease, the dynamics of collective and individual immunity,” says Lukashev.

How the COVID epidemic is developing in Russia. Infographic Society

Evgeny Timakov, the HEAD physician of the clinic "Leader-Medicine", an infectious disease specialist, agrees with this. According to him, the model will be inaccurate if only to trace the movements and crowding of people. “We will need, among other things, data on the mentality, the age of patients, the impact of the virus on each. The accuracy of the model being created will be high only if all these data are taken into account, as well as information about the immunity of the population, risk groups - people who have problems with immunity. Therefore, such work can only be carried out jointly with the Ministry of Health, ”says Timakov. At the same time, he pointed out that it would be necessary to conduct separate studies for each virus, which is practically impossible due to the high cost of such work.

When asked if there are risks for those whose data will be used in the project, Aleksey Pavlov, Business Development DIRECTOR of the Solar JSOC Center for Counteracting Cyber ​​Attacks of RTK-Solar, answered that, most likely, the data for VNIITF is either completely anonymized or carries only a partial sign that does not allow unequivocally identifying a person. “To analyze big data, populations and predict trends, you don’t need personal data of each person – statistics are enough. Therefore, even if the information falls into the hands of intruders, no critical consequences should occur. In any case, for such projects, it is logical to use markers - for example, a social group or a type of disease, rather than specific person data, ”he concluded.