Data scientist on the situation of the coronavirus in Lithuania: It is necessary to stop waiting for life to return to old ways after the second quarantine

“Although we had a record number of new cases this week, it is not yet possible to say whether the situation is getting better or worse. The increase at the beginning of the week and the drop during the holidays are directly related to the increase and decrease in the number of tests.

That’s pretty good to see when comparing the data from the last few weeks by days of the week. When analyzing data for Lithuania, it should always be remembered that the data is several days out of date. Only 70 percent. The tests are summarized within 24 hours, the rest can travel for several days. The whole chain until a new case comes up at the NVSC is about that. Symptoms -> call NVSC / register for test -> take test -> answer (do) test -> summarize test results -> announce case as new. It may take a couple of days from when you sign up for the test until it is done.

The test is taken approximately once a day after collection. Add for up to three days (labs do this by hand). Once the data appears healthy, the NVSC receives it every 3 hours and the case becomes new.

So when new case data is released every morning, it is the actual results of a three-day aggregation. In a process of this type, it is practically impossible to obtain information based on current day data. Only very large changes can be observed and everything else is hidden during the aggregation of different days. Since the delay is approximately three days, the image begins to clear after three days and then problems arise because new cases of NVSC are still an aggregate of three days. From the publicly available data, the pure data for one day is the number of tests and the new positive tests. It is necessary to examine these indicators to see the daily effects, “said the researcher.

According to him, the most interesting effect of the day that I would like to see is a storm of stores before the introduction of restrictions. “Unfortunately, this will probably remain a topic for further scientific studies, because there is a real need to compare the change in distributions over a few days, due to the aforementioned delays, which are also caused by the incubation period distribution. I was assuming we were going to have an effect this week, but as the number of tests decreased over the holiday period, case dynamics and new positive tests are significantly more affected by this factor.

Which is very good and can be seen in the weekly comparison charts. New cases and positive tests have risen steadily in recent weeks as the number of tests has increased in the same way. The amount of testing dropped on Wednesday, but the amount of new positive tests still remained, which is the most suspicious effect of the store storm here, ”the expert wrote.

According to him, it is interesting that testing at mobile points only started to fall on Thursday, but whether it is related to data entry without data or a decrease in test traffic, it will be possible to respond next week. The average proportion of positive studies shows, sadly, that it is too early to speak of an improvement in the situation, as it is constantly increasing every week. It generally increases on holidays and weekends.

“What conclusions can be drawn from this data analysis? The only real conclusion is that so far there is no data on the deterioration or improvement of the situation, which means that we have some stability and now we have to wait to see where things go. While I really want to say something positive, we have to accept the fact that we will be living with covid-19 for a long time and therefore we have to stop expecting everything to improve after undergoing the second quarantine. It is necessary to believe and hope, but it would be rational not to prepare for a marathon, but for a marathon. Now we have a burning house and it needs to be put out, but the house is on a peat bog that has burned well underground, so decisions must focus not only on what to do now, but also on what to do. do in a month, half a year and a year. .

A study by UK researchers on the possible consequences of the new covid-19 strain at https://cmmid.github.io/topics/covid19/uk-novel-variant.html suggests that the situation could be even worse . Its main conclusions are twofold. The most likely reason for the spread of the new strain is its higher infectivity. Even with the most severe restrictions, the UK is facing a spike similar to that of the first quarantine. It can be improved with 2 million vaccines a week. This corresponds to 100,000 vaccinations per week in Lithuania. It is not yet clear if that variety is already in Lithuania, but realistically, it will probably arrive inevitably. The strains of the virus that spread more quickly are quite common: if this is not the case, then another will occur, so it is rational to have a plan for such a case, “wrote the scientist.

It is also important to keep track of the situation. The study mentioned above shows how these questions should be answered in real terms, what is the possible effect of a store storm. With a model that follows the course of the epidemic well, these situations can simply be modeled. To have a model, it is clear that data collection must be of good quality, as data collection errors increase the error of the model.

“It just came to our notice then. Apple’s 12-24 mobility data shows a sharp decline in mobility up to the first level of quarantine. This gives us hope that the restrictions have started to work, which means we will have a faster case. Hopefully this is not a statistical deviation of one day, but the beginning of a new trend, “wrote the researcher.