Under the RADAR

We often talk about some threat sneaking up on us as "coming in under the RADAR", even if we don't know exactly what that means. RADAR was of course the technological magic of WWII which allowed airplanes to be located long before they were seen or heard. (It also allowed the battleship Bismarck to sink the HMS Hood with a single shot. Accurate ranging is particularly important in gunnery.)

Flying close to the ground, where airplanes were hidden in RADAR "shadows" of objects near the RADAR station, was one way to avoid detection. The obvious defensive solution would be to move the RADAR higher, to eliminate "shadows". This runs into a second problem "ground clutter", returns from objects near the ground which are not aircraft. Around 1950 the technique of "clutter mapping" was introduced to eliminate returns which did not change, as a return from an aircraft in flight would change. This will still fail in the case of moving objects like the branches of trees in a wind, or waves on a body of water. The problem of separating returns which signify threats from those that do not still presents a problem. And, it is still possible for aircraft to sneak in "under the RADAR" if they know exactly how to do it, though today this may require active countermeasures, possibly by other aircraft.

In the context of infectious disease this translates into pathogens which exploit weaknesses in immune defenses to avoid detection. The defensive problem is again a matter of separating real threats from apparent or imaginary ones. A living body does not exist in isolation. A human, for example, must expect a diverse intestinal flora to colonize the gut. The ability to digest certain foods depends on these to break down food items into small molecules which can easily pass through the walls of the digestive track. As omnivores we have less in this regard than herbivores, but more than obligate carnivores like cats. Every mucus membrane has its own ecology of microbes, and there is a difference between healthy and unhealthy ecologies. The problem for the immune system is immensely complicated by the need to tolerate and protect beneficial microbes. The immune system has to do a lot of "clutter mapping".

It is not enough to tell "self" from "other", as in older descriptions of immune response. The context in which "other" things are found is crucial, as in the case of pathogens and damage to "self" tissues. You can tolerate e. coli doing its job in the gut; you cannot tolerate it attacking cells in the blood stream. A second aspect is behavior over time, like those radar images which move.

This is where I get into a simple mathematical analogy: the distinction between (quasi) linear and exponential sequences. Don't take this too literally. What I'm getting at is a crucial distinction between living things and inanimate matter -- replication. I'm going to present some sequences of numbers with the implicit idea these are time sequences. For simplicity these will be exact sequences without random variation.

The classic linear sequence is counting: 1, 2, 3, 4, 5, 6, 7...

The rule generating this sequence is to add 1 to the previous item in the sequence.

The classic exponential sequence is this: 1, 2, 4, 8, 16...

The rule generating this is to multiply the previous item by 2. I think the connection with replication of living things is obvious.

If this were all we had to deal with things would indeed be simple. It takes a good bit of mathematical sophistication to realize that the following sequence is also exponential, but with a different base: 1, 1, 2, 3, 5, 8, 13...

(This is called a Fibonacci sequence, and can be obtained from the integer part of successive powers of the real number Phi = (1+Sqrt(5))/2 )

We can get into difficulty with far less sophistication.

This is also a linear sequence: 1, 1.1, 1.2, 1.3, 1.4, 1.5...

The rule here is to add 0.1 to the previous item.

This is also an exponential sequence: 1, 1.1, 1.21, 1.331, 1.4641...

The rule here is to multiply the previous item by 1.1.

What should immediately jump out at you is the similarity of the last two sequences. Would you predict that the values in the last sequence would pass the first sequence, and just keep going up? Not if your data for the first few items was uncertain.

Now, in biological terms, the important distinction is between things that vary slowly and things that behave like living threats. Given time, an exponential increase, even with a small multiplier, will pass any (quasi) linear time sequence. If you wait long enough it will pass any threshold you set as a "tripwire" value. The problem with waiting is that the organism you are protecting may be doomed by the time you discover a threat this way.

Current standards for detecting infectious disease do a good job for things that multiply like 1, 2, 4, 8... They do poorly for things that behave like that last exponential sequence. What we end up with are diseases like cancer, which may have a very long prodromal period lasting years, but then begin an obvious exponential increase which is very hard to stop. The easy interventions to stop this pathological process would have to be made at a time when current consensus says no disease is present. In fact the disease called cancer may not be present at all in the early pathology. More and more it looks like the cancer is a runaway process caused by an imperfect response to a chronic condition, which may well be infection. The response can hold the causative agent at bay for a long time, but ultimately generates a more severe threat.

For infectious diseases of the CNS we have a larger problem, since the CNS is largely inaccessible, and much of the pathology is only apparent at autopsy. We assume pathology inside the CNS will generate changes in blood because this assumption is convenient, not because it is true, except in the vacuous sense of being true by definition. Analysis of cerebrospinal fluid is much rarer than blood tests -- for one reason because it requires a lumbar puncture.

The above numerical analogies are not the worst possible cases. After reading a paper on a possible infectious cause of Alzheimer's disease by spirochetes of genus borrelia or treponema I've decided to add an example to my discussion of slow exponential increases to illustrate a really slow exponential increase.

Consider the following linear progression:

1.0, 1.01, 1.02, 1.03, 1.04, 1.05...

The rule here is to add 0.01 to the previous item.

Now consider the following exponential progression:

1.0, 1.01, 1.0201, 1.030301, 1.04060401, 1.0510100501...

The rule here is to multiply the previous item by 1.01.

Round this to two decimal places, and the later increase beyond any linear increase will come as a complete surprise. In a medical context, this increase will be called a separate disease.

Here's what the 51st numbers in each sequence will look like:
linear 1.50
exponential 1.64463...

Here's what the 101st numbers in each sequence will look like:
linear 2.0
exponential 2.70481382942...

Here's what the 201st numbers in each sequence will look like:
linear 3.0
exponential 7.31601785183...

Here's what the 301st numbers in each sequence will look like:
linear 4.0
exponential 19.7884662619...

Researchers looking for a separate event triggering the apparent sudden change will find nothing. In this case, nothing should be an important clue that you need to revisit starting assumptions.

Note: I'm treating the first item in those sequences as the zeroth application of the given rule, which is why I'm saying 51st instead of 50th, etc.

I'll add another observation: the pathological processes which are easiest for doctors to detect are also the easiest for the immune system to detect. In fact many medical signs and symptoms are caused by immune response instead of the pathogen itself. This means gaps in medical practice concerning infectious disease tend to match natural vulnerabilities. Many people treated for infectious disease by doctors would have recovered without treatment. This is not reason for complacency. It points instead to a likely source of failure in medical practice. If you don't look for these gaps I can guarantee you will not find them.

Our emphasis on pathogens with rapid replication also contributes to the fallacy of always assuming a single pathogen per infectious disease. If one species dominates all others in patients and culture through more rapid replication it will be the only one you have to deal with. When replication is slower you have to go much further down the pathological process to have such a clear winner. If the disease compromises immune competence you may never get such a single pathogen.

In one Alzheimer's Disease (AD) patient brain examined in that research above they found six different species of treponema, none of which was t. pallidum. The patient did not have neurosyphilis. Tests for borrelia burghdorferi would also fail, because this was not neuroborreliosis. Only by performing examinations and tests which were neutral with respect to precise species of spirochete could they reach a conclusion that the disease was likely caused by infection. Particularly striking was the observation that neutral tests for spirochetes identified regions of the brain with characteristic pathology of AD. Emphasis on a single pathogen per disease has produced diagnostic fragmentation of a common pathology. You can never get the kind of correlation desired from any test for specific species. Finding spirochetes in the brain is, however, a very strong sign. That research found them in 90.1% of 680 patients, and 0% in 185 controls. Measures of statistical significance are way beyond ordinary medical standards.


"the pathological processes which are easiest for doctors to detect are also the easiest for the immune system to detect."

And if the easy ones aren't found, well, then the patient must not be sick, so we don't need to investigate any further. Besides, their 10 minutes are up. Next claim generator, err, patient please.
Hi jimells, your description only covers the immediate interaction between doctor and patient without considering the effects it has on insurance, bureaucracy and "evidence-based medicine". Amplification of errors by social networks is like concentration of toxins in food webs.

If you start with the consensus, just as a random example, that there is no Lyme disease in Florida, few doctors will test for it or report it. This will lead to evidence-based medicine showing that the presence of the same species of ticks (ixodes scapularis) and white-tailed deer mysteriously fails to produce Lyme disease in the corresponding human population for unexplained reasons. If you define any case with the visible diagnostic signs of Lyme disease as STARI because of location alone you can accentuate this diagnostic bias.

Evidence-based medicine could easily show that all patients are right-handed. Careful examination of those who claim to be left-handed would reveal no physical impairment of the right hand. This, plus strong statistical evidence, would lead to the conclusion that all those claiming to be left-handed have mental problems.
In the paper by Miklossy above, I was, at first, unaware of the meaning of the reference to "criteria of Koch and Hill". Koch is well-known to microbiologists, but I did not initially connect that Hill with Bradford Hill's work on Bayesian methods in epidemiology. Here's a reference on these statistical methods in typical epidemiological research where past data are inadequate for a pure frequency-based approach.
Controversy about the philosophical basis should not obscure the fact that these methods worked in finding risk factors for lung cancer and heart disease at a time when the evidence was inadequate for prevailing statistical orthodoxy. Using such methods, the suspicion of chronic infection by spirochetes as a cause of dementia becomes very compelling.

Blog entry information

Last update

More entries in User Blogs

More entries from anciendaze