Why not the week of India's expected conditions in hospitals: 'It was an advantage'

Why not the week of India’s expected conditions in hospitals: ‘It was an advantage’

It will be this week that, according to RIVM Models, things could totally go wrong in the Netherlands, with Indian cases in hospital. But despite the squeaking and creaking of the ICU, everything worked fine. Question by Dutch RIVM designer and designer Jacco Wallinga: Didn’t the models exaggerate things a little?

Two thousand people, all of them on a ventilator. Or, if things are not going well, maybe six thousand, when there is only a place for about 1,500 in intensive care. This was the grim prospect that had constantly loomed over the horizon of RIVM predictions in recent months.

This week will be the day. The “third wave” will reach its climax, with as many as thousands of Covid-19 patients in the IC. Only: Fortunately, there were not more than about 840. Together with patients who are not infected with the virus, enough to put a heavy burden on hospitals. But also as if they had anticipated a typhoon, after which there was nothing but heavy rain.

That’s how those expectations go, says RIVM chief designer Jaco Walinja on the phone. The point he should go on to explain: In his world, a rainstorm and a hurricane are equal probabilities, “only uncertainty exists,” he says. Some people think that the range of uncertainty in our predictions is like statistical hype. If you try with more effort, you can filter out the noise and you can capture the real signal with more accuracy.

But with RIVM models it’s a lot like a dice roll: It can go that way, but also. RIVM rolls the dice 200 times for each prediction, to get an idea of ​​what to imagine. This is how we reach those fans. There are usually several model results below. The fan becomes thinner towards the top.

See also  LEGO pays tribute to science with a free set

However, for months on end, your department has shown us graphs of thousands of ICU patients. Don’t you think yourself: This sounds nonsense to me?

I think these outliers are typical of all biological models, including infectious diseases. This is because infectious diseases tend to grow exponentially, which continues at a tremendous speed and for a long time, until the proportion of people exposed is very small. This tendency to explode is very much present in the system. And if you look at India or Brazil, it really is. Every now and then things get worse – and then get terribly wrong.

In our country, we have seen numbers increase again since the beginning of February. What we saw: The road has begun, there is a beginning of exponential growth. And there was absolutely no information about when this exponential growth would stop. This gives you a very large set of possibilities up.

Until Easter, early April. Then it became apparent that the exponential growth was stagnant. For the first time, the model indicated: Wait, this is no longer compatible with exponential growth, we must be close to the peak.

Country Viewer, this is the best way to contact him. 52-year-old Jacco Wallinga, a thoughtful drafting scientist and professor of sports infectious disease modeling in Leiden, who has gained a major standing in the sciences through his work on R-numbers and modeling of grafting effects.

Then the aura came. His calculations were that in the past year and a half, it had often made the difference between opening and closing shops, bars and schools, or extending curfews or not. His colorful futuristic charts, which put politics on the brink of last March and December last year: Brace Yourself, Golf is on the way.

With self-mockery: “Let me put it this way, I didn’t start working for RIVM to become kind of a famous Dutchman.”

Since the start of the third wave, your models have said this may turn black. But it couldn’t be that bad either. Are you not actually saying: It can go in any direction, we have no idea?

‘I do not think so. In mid-December, when the British Substitute introduced itself, we suddenly saw: The numbers could soon rise again and very quickly. It was hard to say exactly where we were going to end up. But it was clear that additional measures were required, which were more stringent. Our message was: Be very careful. This is more than: We don’t know.

Ultimately, the real numbers stayed below the average for your models. It gives the feeling that your models were too pessimistic.

One way or another, many people look at our photos this way: as if this average is what to do. But this is not the message we want to send. The median isn’t the most likely outcome, it’s actually the median: 50 percent of the simulations are higher and 50 percent less. So this is a different creature than the most reasonable value.

In addition, we mainly focus on forecasts 14 days in advance. From this we say: we are absolutely sure of this. In that period, the average turned out to be a very good indicator. In the future, however, it becomes more speculative. Often this includes information that we cannot supervise, people’s behavior or about vaccinations.

How do we then know if we really averted disaster with additional measures?

“Look at the numbers: They were close, even with the most stringent measures taken. A scenario in which the number of admissions to the medical center consistently exceeds hospital capacity has been a realistic option, and it is more realistic than ever last week.

We have now managed to push the peak ahead of us. It is now in May, and now we’re benefiting from the seasonal effect and vaccinations. I guess we just make it now. Or not, it remains to be seen.

Because? Has the third wave really ended?

As now, you just have to take into account that we could still have more IC pulls. But this opportunity diminishes. Out of the 200 simulations we run, most indicate that we are at the top, although there are a few that still point towards the top. Just that there is less and less. Hopefully this real peak is totally gone from the models next week.

How do RIVM models actually work?

RIVM models show how people relate to one another, broken down by age and type of activity – information that RIVM has collected for years. The virus is then transmitted in some kind of computer game through a “communication matrix” to study the number of people who contract the disease. With each simulation, the initial conditions are slightly modified, with few or fewer contacts, and transmission of the virus slightly more or less, in order to avoid accidental fluctuations. In this way, after two hundred cycles, a “fan” of lines for the future is created.

This, of course, is the simple explanation: In real life, models also encode things like the effect of season, vaccinations and of course, the effect of actions. Specific adjustments can also be made to the contact matrix: for example, it is possible to simulate what happens if the number of contacts at work halves, or if schools are closed.

Leave a Reply

Your email address will not be published. Required fields are marked *