Numbers and testing as a weapon against the pandemic
The subject of another Neuron Club Online was numbers and testing as weapons against the pandemic. The Neuron Club is an exclusive discussion platform for patrons and leading Czech scientists, where they are given space for their reflections and independent ideas. Look with us behind the curtain at the most interesting things said at this online meeting of scientists and patrons.
There are currently mathematical models for almost every known infectious disease. They make it possible to describe and, above all, to estimate the parameters of an infection from the available data shortly after the outbreak begins. Thanks to them, society is then able to adopt suitable measures and confront the disease effectively.
Living non-living things, bad news wrapped in protein… meet the viruses
The Latin word virus means poison or slime; it took the electron microscope to show that these are particles of outstanding geometric beauty. They are found all around and inside us: one litre of seawater contains more virus particles than there are people on the planet; on our bodies and in our gut we have 10 times more bacteria than cells, and 10 times more viruses than those bacteria. We routinely take viruses in with water and food… but most of them do us no harm.
The viruses that cause disease are the curiosity. A virus's evolutionary interest is a) to multiply and b) to spread to further hosts. Outside a host cell, a virus in the environment shows no signs of life. It comes to life only at the moment it infects its host, that is, reprograms, suppresses or halts the processes in the original cell and produces only new virus particles (an analogy with computer viruses, which do the same). If a virus kills its host, it dies itself — so that is not its primary aim; a clever virus wants to survive. "COVID-19, though, behaves oddly in this respect and may yet surprise us."
Before the SARS epidemic, coronaviruses were of no interest, which changed after SARS and MERS. COVID-19 is the third major coronavirus epidemic. The trouble with the COVID-19 epidemic is that, unlike SARS and MERS, a great many people have a mild course of the disease or none at all, and so can spread the infection more effectively.
Beware of interpreting the results
The parameters of the tests are unclear. Even with a relatively large sample of the population tested (28,000, largely a representative sample), it is not possible to interpret future results clearly, because we do not know the test's parameters. Its accuracy is between 91% and 97%. So with any result below 9% we cannot rule out zero prevalence (9% may be a false positive or false negative result).
The presence of antibodies may be a problem. A large number of people who have had a mild infection have no antibodies at all, so the tests will not detect them anyway. The question therefore remains what the point of such testing is, "because colossal decisions will then be made on the basis of it." "The whole country is watching one dramatically imprecise experiment." It is therefore essential that decision-makers recognise these limits and do not base future steps on such unclear results. "It is just one more piece of the information jigsaw, not a picture of reality."
Numbers as a weapon?
A model is a simplified version of reality created for a particular purpose. When constructing one we have to determine in advance what it is to serve and what questions it is to answer (a paper model of an aeroplane versus a flight simulator). Every epidemic has its mathematical model, and there are a great many of them — each author asked a different question. All of them share a common basis from the 1920s: SIR models, which model transmission between the categories of recovered, infected and susceptible.
A mathematical model of the COVID-19 epidemic is currently being created, with the aim of describing in detail the effect of the measures adopted and predicting further developments. It consists of four layers:
a) the epidemic layer — mapping how individuals become infected, how infectious they are, whether they have symptoms and so on,
b) the hospital layer — how many patients are in an ordinary bed, in intensive care, on ventilators,
c) the quarantine layer — contact tracing;
d) the testing layer — we need current data in order to calibrate the model correctly.
Predictions are hard, especially about the future
The problem with epidemiological models is that they are not rich enough to describe the real number of variants of situations in the real world, the role of decision-makers, the number of asymptomatic people and so on. They are very useful, but it is important to accept the ignorance that remains after them. "Worse than not knowing is false knowledge. And making decisions on that basis, because too much is at stake."
A smart solution
If we cannot predict, is letting the disease run through the population a possible route? If the epidemic is let loose without restriction, it will affect 5–8 million people, of whom 40,000–100,000 will die, the health system will be overwhelmed, and moreover it is not clear whether people will develop sufficient antibodies. So what is to be done?
Governments have essentially three ways of dealing with the situation: closing borders; closing public spaces; testing and tracing.
Tracing and smart quarantine as the way out
It is a cheaper alternative to repeatedly opening and closing schools, businesses and so on. It is the only proven solution that works elsewhere in the world (South Korea, for example), unlike letting the population acquire immunity by infection, about which we have no data.
Effectively breaking the infection network — R for COVID-19 = about 3 (one person infected roughly three others). But it is important to realise that the distribution is very uneven: most people infect no one, while superspreaders infect 15 people or more. That is the virus's weakness — if we deploy tracing intelligently, we can eliminate superspreaders from the network and so break the virus's contact chain effectively.
