Does Chaos Theory apply to our illness?

[Ecoclimber]

Maybe all that's needed is to stomp out that stupid butterfly! I hear MIT is working on something with their supercomputers.

 

[user9876]

Definitely. Its also why, in my opinion, many interventions have temporary effect. You push the system. You feel better. But its a stable system and pushes back. Medication "stops" working. Actually its still working, you have just moved to a slightly different stable state.

The way I am thinking is there are a number of dynamic processes involved with the basic management of the body (just don't ask me to list them). I'm assuming that they will get some control signals from the brain that help them adapt to the current state of the body. Generally they would be kept in an equilibrium by such action or pushed into an alternative equilrium which is suitable for managing a current change (for example during an infection the immune system will have various response loops). So that would be normal working which with feedback loops is likely to be non-linear in nature given a number of inputs.

So then if we follow something like some of the recent brain scanning work where if I remember correctly microglia were active in areas of the brain associated with controlling parts of the basic body functions. Then say a delayed incoming signal or a delayed or slightly wrong response could break the normal feedback loop which would maintain the system in a region where there is an equilibrium but push one or more parts of the body's management processes into a chaotic region and hence lead to a time or worsening symptoms etc but also perhaps a high degree of variation hence when measuring we may never get consistent measurements even within one person let alone across groups.

If we look at weather forecasting (which done through a complex series of non-linear equations with feedback) then we have stable weather states and unstable and unpredictable ones. What the forecasters do is they run their models with slight variations of the input parameters - often they get much the same predictions and hence can safely predict but sometimes they get a situation where the predictions are wildly different and they known they can not predict what will happen next. Although they still make a best guess since thats what people want - but they know it is highly uncertain.

So back to the body. If we see this in a similar way as a non-linear system has it evolved so that processes move between stable states but based on certain parts of the system always working. If we were to change say the speed of response of increase a loss rate on signals within one of these loops could we get into a situation where the behaviour of a process becomes unpredictable and erratic or perhaps this situation would occur when there are other stimuli such as exertion.

If the systems in the body that people are trying to measure are pushed into a chaotic space and hence unpredictable this could have a huge impact on the search for biomarkers as it may be that they change often and in unpredictable ways between people. Hence contradictory results from various groups.

It is one of the reasons I think it would be interesting to do a very detailed study on a few patients with very regular testing to sample what is happening over time. The other is generally if there is a dynamic system it may have time varying (or stimulus varying) behavior which is predictable but if we measure at a given point without knowledge of the variations then we can get a false picture. Like if we always look out of the window at the same time we wouldn't know that light varies over the course of day and night, This could be another reason for variations in measurements between research groups since there experiments may always measure a state at a given time or a given pre-exertion level but others have different times and exertion levels etc. Hence they all get a different view.

My feeling is that we don't have enough information about the way the body works to do any sort of dynamic modelling of systems to try to explore this from a mathematical angle. But I really don't have much knowledge or understanding of biological systems.

 

[Marco]

So back to the body. If we see this in a similar way as a non-linear system has it evolved so that processes move between stable states but based on certain parts of the system always working. If we were to change say the speed of response of increase a loss rate on signals within one of these loops could we get into a situation where the behaviour of a process becomes unpredictable and erratic or perhaps this situation would occur when there are other stimuli such as exertion.

That sounds like a similar process I described some time ago in a blog on complex regional pain syndrome where various autonomic etc symptoms develop well away from the initial injury site :

(my words) :

First, an injury in the periphery (body) activates first order nociceptive (pain sensing) neurons to release the signaling cytokine CCL21, which then triggers secondary neuroinflammation in the thalamus, creating a ‘central generator of neuropathic pain’.

Neuroinflammation or neuroinflammatory tracks that spread through the nervous system alter neurotransmission and coding, causing higher order functional changes in the nervous system including, perhaps, deficits in higher level brain tasks such as executive functioning.

Read more: From Chronic Regional Pain Syndrome to Fibromyalgia to ME/CFS? The ‘Spreading Neuroinflammation’ Model http://www.cortjohnson.org/blog/201...lgia-mecfs-spreading-neuroinflammation-model/

 

[Dr.Patient]

The way I am thinking is there are a number of dynamic processes involved with the basic management of the body (just don't ask me to list them). I'm assuming that they will get some control signals from the brain that help them adapt to the current state of the body. Generally they would be kept in an equilibrium by such action or pushed into an alternative equilrium which is suitable for managing a current change (for example during an infection the immune system will have various response loops). So that would be normal working which with feedback loops is likely to be non-linear in nature given a number of inputs.

So then if we follow something like some of the recent brain scanning work where if I remember correctly microglia were active in areas of the brain associated with controlling parts of the basic body functions. Then say a delayed incoming signal or a delayed or slightly wrong response could break the normal feedback loop which would maintain the system in a region where there is an equilibrium but push one or more parts of the body's management processes into a chaotic region and hence lead to a time or worsening symptoms etc but also perhaps a high degree of variation hence when measuring we may never get consistent measurements even within one person let alone across groups.

If we look at weather forecasting (which done through a complex series of non-linear equations with feedback) then we have stable weather states and unstable and unpredictable ones. What the forecasters do is they run their models with slight variations of the input parameters - often they get much the same predictions and hence can safely predict but sometimes they get a situation where the predictions are wildly different and they known they can not predict what will happen next. Although they still make a best guess since thats what people want - but they know it is highly uncertain.

So back to the body. If we see this in a similar way as a non-linear system has it evolved so that processes move between stable states but based on certain parts of the system always working. If we were to change say the speed of response of increase a loss rate on signals within one of these loops could we get into a situation where the behaviour of a process becomes unpredictable and erratic or perhaps this situation would occur when there are other stimuli such as exertion.

If the systems in the body that people are trying to measure are pushed into a chaotic space and hence unpredictable this could have a huge impact on the search for biomarkers as it may be that they change often and in unpredictable ways between people. Hence contradictory results from various groups.

It is one of the reasons I think it would be interesting to do a very detailed study on a few patients with very regular testing to sample what is happening over time. The other is generally if there is a dynamic system it may have time varying (or stimulus varying) behavior which is predictable but if we measure at a given point without knowledge of the variations then we can get a false picture. Like if we always look out of the window at the same time we wouldn't know that light varies over the course of day and night, This could be another reason for variations in measurements between research groups since there experiments may always measure a state at a given time or a given pre-exertion level but others have different times and exertion levels etc. Hence they all get a different view.

My feeling is that we don't have enough information about the way the body works to do any sort of dynamic modelling of systems to try to explore this from a mathematical angle. But I really don't have much knowledge or understanding of biological systems.

Weather predictions become more and more accurate as the time gets closer and closer to a given day. As such, any research being done should be done on the sickest, bedridden patients--where hopefully, the variations are minimal.

There are hundreds, if not thousands, of parameters that are available that can be used in dynamic modeling with this illness--it's just that there is no money, or the interest. The only enticement is that this illness is wide open, and if anybody wants to make a name, or donate money to a not well known illness, are they willing to do it...

 

[anciendaze]

You don't need thousands of parameters to generate or model complex dynamic behavior. There is existing work on reconstructing attractors in phase space from a time series of a single measurable parameter. I was aware of this work in the 1980s because I was working in the simulation business, and it offered a window into those parts of systems considered "black boxes". The best medical work on time series comes from cardiology, and some implanted defibrillators now use software based on nonlinear dynamics. (Though cardiologists who implant these do not need to understand the theory behind them.)

At the time I was somewhat interested in application to medical mysteries, but didn't know it had any relevance to myself. (I had been told what I had was "a personal problem" rather than a medical problem.) I talked to physicists who had tried to interest medical professionals. What I got back was a story about an attempt to find the underlying dynamics for the clinical sign of "schizophrenic gaze" (optic nystagmus). This is not only present in many schizophrenics, but also in some close relatives without psychiatric diagnoses.

They took a movie film of healthy people tracking a target moving on a screen, and one of a schizophrenic performing the same task. They then digitized eye positions to get their time series. While most researchers expected to get a huge mess indicating random behavior, they found that most of their reconstructed attractors only required three dimensions, the minimum needed for chaotic behavior. This indicated deterministic chaos with a small random component.

The natural inference from this was that the clinical sign was based on a fairly simple neurological problem. They couldn't point to a particular nerve and give the Latin name, but they could say they had objective evidence this did not involve large parts of the brain. They presented this at a conference, and were shocked to hear speaker after speaker rise to denounce their work, telling them they didn't begin to understand how complicated visual tracking was even in the absence of schizophrenia. These physicists felt they had solid results, but the profession was not willing to listen to them.

It was in discussing this later with doctors that I came to realize most doctors equate "abstract" with "vague" and "theory" with "speculation". Most simply have no experience with rigorous abstraction or mathematical modeling.

As an example of how such experience can cut through layers of complication and obfuscation introduced by organizations, consider this story about the dynamics of the Space Shuttle main engine turbopumps. I sat in on a talk by a consultant hired to help determine the optimal service interval for those turbopumps based on analyzing data returned by telemetry during launch. The original idea was that those pumps might go many flights before it was necessary to rebuild them. (Remember that the Shuttle was supposed to reduce costs by being reusable?) In the overheads for his talk showing the spectrum of vibration I noticed the classic evidence that the engine was producing subharmonics, something which only happens in non-linear systems. What is more, as operation progressed those separate subharmonics started to merge. To me this was definitive, even if the data had not been presented in the way I preferred.

I couldn't get the speaker to make any public statements about his recommendations, but it was clear to me that the optimum number of flights between rebuilding turbopumps was one, and this was only because it wasn't possible to rebuild those pumps in flight. Fortunately, those in charge of the Shuttle program also realized this, and an engine explosion never caused loss of an Orbiter.

 

[anciendaze]

Here's an earlier blog post of mine which may (or may not) appeal to readers of this thread. In passing, it references the role of chaos in a purely mechanical context most people don't even imagine.

Part of what I'm trying to convey is the presence of weak chaos in systems which tolerate limited random variation. This is common in healthy living organisms, but, as I'm trying to illustrate, it is not a purely biological characteristic. (It is even present in the long-term dynamics of the Solar system. )

The interaction of conical wheel bearing surfaces with curved tops of rails can be modeled mathematically. This combination was found empirically long before the mathematical concept of deterministic chaos existed. The result is that the trucks on a rail car will "hunt" even on perfectly smooth track where there is no variation. Various attempts to eliminate this completely, for the comfort of passengers, came to nothing. You can even hear the characteristic sound of this process on modern high-speed trains on welded rails, where variation is much better controlled. This behavior is necessary for the train to tolerate tiny variations in rail spacing, heading and loading. Without it trains would frequently fail to negotiate curves, and tracks would need constant repair. On high-speed trains, limits on the extent of this behavior place a critical speed limit beyond which safe operation cannot be guaranteed. To get beyond that you need to eliminate wheels.

Incidentally, there is a toy which demonstrates that this particular example of mechanical adaptive behavior does not require a complicated system with many components. If this needed precision machining it would be too expensive to be a cheap toy. Track cost was a major factor in the development of railroads. The toy is called Wheelo.

Common medical assumptions about stable values and fully predictable behavior being good, and unpredictable behavior being random and bad, are at odds with reality. All biological control systems which have been carefully investigated show similar weakly chaotic behavior around fairly stable values. In fact the change in heart rate variability to purely predictable is a very bad sign of impending doom.

We have many problems indicating a disturbance in physiological dynamics, but eliminating chaos is not a real option. Understanding those dynamics should lead to evidence of processes which are totally missing in our current physiological models. Current emphasis on levels of particular biochemicals, with no corresponding information on rates, doesn't tell us much of anything about dynamics. This collapses the phase space into a mess with half the number of dimensions.

Full understanding of the problem could take a very long time, but I'm convinced we could pull out useful information about such things as posture and gait disturbances, or regulation of blood pressure, in short order.

What I cannot countenance is more of the same biomedical research which has gone nowhere. For historical reasons medical standards of practice were long designed to exclude such inconvenient illnesses to treat from medical practice. Times are changing.

 

[anciendaze]

When I wrote the above originally I had not seen a very relevant discussion here on PR.

Here's an example of simple physiological dynamics which could be put to immediate use. It turned up in a thread started by Jodie1980.

She meets the requirements for a diagnosis of POTS, though there could be some serious heart problems beneath that. I think the telling thing is that she did this herself. She has already been to her GP, and has investigated specialists she might go to. She is on waiting lists.

There was a local case in which a young woman diagnosed with POTS committed suicide. The circumstances were taken as evidence of "hidden depression". Naturally, no one assumed medical neglect played any role. (I've removed an explicit link to news reports I used in a private message to avoid causing that family further unnecessary pain.)

I contend she did not have depression, per se. She had a recognized autonomic dysfunction which was labeled without being treated. Nobody seems to have told her to limit time spent upright, or to be careful about fluid and electrolyte balance. Nobody warned her about making very bad decisions when her brain was starved for oxygenated blood. You don't send pilots to psychiatrists if they make bad decisions due to hypoxia, you correct the problem.

This is a problem which can be addressed with available treatments, even if it cannot be cured. Diagnosis only requires using existing equipment and thinking.

Why are patients with such severe examples of dysautonomia having to test themselves, and go on waiting lists?

This is not the most extreme example of malpractice following official guidelines. There were newspaper reports concerns a woman in the U.K. who had an illness reported as sudden-onset ME/CFS following a probable viral illness. The strange thing here is actual paralysis, which is not typical. Even extreme weakness which totally prevents patients from standing is not typical of ME/CFS. Both conditions do show up in channelopathies typically involving potassium or sodium channels. The fact that the paralysis or weakness comes and goes is often used to classify the problem as "mental" (catatonia). Anyone who has seen an EKG made while such a patient is having such an attack will drop psychological explanations. The condition is said to be rare, and about half genetic.

The problem with current healthcare in the U.K. is that "CFS" is considered a mental health issue which can be overcome by counseling and exercise. Even though there was "specialist care" for that condition, they had no way to deal with patients who were actually paralyzed. This meant she would be transferred to a mental hospital locked ward when she could no longer be cared for at home. Once she figured this out, suicide was predictable.

Even at this late stage we don't see evidence anyone is officially considering possible misdiagnosis.

 

[anciendaze]

While on the subject of railroads, I'll put in a plug for an earlier blog post of mine concerning trainwrecks. On the subject of neglect of dynamics, and the assumption variation must be random, I'll mention the blog post on aleatoric medicine.

 

[anciendaze]

I want to emphasize that what I've said about the importance of dynamics in physiology, and medical neglect of same, is not wild speculation. Virtually every aspect which has been investigated in detail shows the existence and importance of variations which are commonly dismissed as random.

I've made another blog post about the level of ignorance concerning immune function in current medicine. This is not to say that research literature on immunology is lacking. There is a plethora of material, most of which is of very limited use to physicians treating individuals. What is most strikingly missing is any conception of dynamic response. This despite the common observations of such signs of immune response as variation in fever.

Now I'm going to go beyond the safe material I can back with solid references, and make a speculative hypothesis. I don't promise to produce a unified theory of ME/CFS from this, or even to defend this idea against criticism. I'm just putting it out there for consideration.

When I went into literature on such subjects as mitochondrial function, oxidative stress, reactive oxygen species, etc., which appear to be central to our problems, I frequently found the same terms appearing in research on reperfusion injury, as happens in strokes. Nothing limits such damage to nerves or the brain. It can feature prominently in heart attacks.

Numerous reports of problems with the gut also connect with hypoperfusion, as is demonstrated in cases of heart failure. (I'm indebted to Paul Cheney for this insight.) A normal part of the "flight or fight" response is to reduce blood flow to the gut to allow more oxygen and glucose to reach muscles. Dysautonomia obviously works against us in that case.

One other insight from Cheney was that immune function is strongly dependent on cardiac output. This makes sense because immune cells must reach affected tissues, and chemical signals from affected regions must reach other immune cells, before much significant (or carefully directed) immune activity can take place. The variable least considered in medicine seems to be the rate at which this happens, or, equivalently, the transit time for immune cells in the bloodstream.

In extreme cases of circulatory problems no one is particularly surprised if extremities consistently receiving reduced blood flow become infected, (as with gangrene). What seems to be missing is any awareness of dynamics of blood supply, even in cases as extreme as patients on the edge of syncope whenever they stand up.

I'm wondering if repeated episodic hypoperfusion, (and not only in the brain,) is behind a great many examples of cumulative damage to tissues. Allowing commensal bacteria to penetrate more deeply into gut linings would obviously result in an immune response like polyclonal B-cell expansion, and might result in autoantibodies against proteins (like cardiolipin) that resemble those found in bacterial membranes. Conditions like asthma and migraine show that this kind of behavior at boundaries of the immune system is not unique to gastrointestinal mucosa. Episodic hypoperfusion is observed all the time, and generally ignored on the grounds that the process will be naturally corrected.

Just how far is this from being the exclusive problem of a neglected minority of patients? Consider that about 1/3 of all trips to emergency departments involve syncope, and hypoperfusion is a major cause of syncope.

 

[anciendaze]

I was hoping to get some feedback, positive or negative, on that last post. The silence I've heard, even with a number of people following my posts, tells me there was substantial non-comprehension. I'll make another attempt today.

What I've said previously is that you can't eliminate chaos from our bodies because weak chaos is a property of all robustly-stable systems in unpredictable environments. Healthy people exhibit weak chaos in such things as heart rate, regulation of blood pressure, gait, breathing, etc. In the largely uncharted territory of the dynamics of immune response, I predict that chaotic dynamics will turn up everywhere.

Popular accounts of chaotic dynamics tend to concentrate on a single feature: sensitive dependence on initial conditions. People also carry intellectual baggage with them into this field, thinking this is always bad.

In the technical literature on chaotic dynamics you will find everything discussed in terms of a phase space. This is a very powerful abstraction introduced in the 19th century. A single point in phase space of a purely deterministic system carries all the information needed to predict future behavior. Half the coordinates of a point in phase space are familiar ones like position in x,y or z; half are "momenta" telling you how rapidly those coordinates are changing at the moment. This means you can talk about that behavior as an "orbit" (trajectory) in phase space, which turns this into a kind of geometric problem involving the line traced by a moving point. (Poincare did this in trying to answer the question of stability of the Solar system. He couldn't say for sure that it was stable. We still can't.)

That sensitive dependence on initial conditions is only one aspect which causes a region of phase space to be labeled chaotic. Another requirement is that the region have a dense network of periodic "orbits". Predictable behavior is not entirely excluded, but it is very easy to switch from one kind of behavior to another. (Check for examples of period-doubling. These don't have to be unusual. This happens in dripping faucets as the rate of flow increases, as I've tested. There are even scientific papers on dripping faucets.) The third common requirement is called ergodicity, which I won't try to explain in detail. The central concept of this is that a chaotic trajectory ("orbit") in a region will pass close to every point in that region. From the standpoint of switching control from one behavior to another this is valuable, if problematic. The study of how to control chaotic behavior even in machines is still in its infancy. Nature got there long before us.

Now if you take the conventional medical approach you will label everything like this "random", often forgetting that random essentially means "we don't know what will happen next". This will cause doctors to ignore what goes on in patients short of some clinical sign predicting imminent death, which calls for intervention, even if the odds are against success. The truth is that we don't understand what is happening in patients who are sick, but likely to fully recover. Recovery can take place in hours, days, weeks or months, and we have little basis on which to predict which patients will take which times. The possibility that there are previously-healthy patients who will remain sick for years, without dying and providing convenient specimens for pathologists, is resolutely rejected. These patients are anomalies to current medical theories (in the terminology of Thomas Kuhn) and are treated with the scorn typically reserved for anomalies you can't either predict or change. We talk about "healthcare" without having much of a clue about the causes of health in those who recover from short-term illness.

Those of us who remain in "recovery" states for years are subject to exaggerated forms of sensitivity to perturbations, and exhibit many kinds of changing responses, as our damaged bodies try to deal with an unsatisfactory situation. I am far from the first to note the similarity with problems seen in normal elderly people.

One typical predictor of cognitive decline in the elderly is repeated syncope (fainting). This is pretty clearly associated with control of blood pressure and hypoperfusion of parts of the brain. You can't expect to think very well if your brain isn't getting oxygen and glucose. There is no question such episodes are not good for the brain. Other organs can also be damaged by localized reductions in blood flow, though the results, except in the case of the heart, are less dramatic.

Another sign of dysregulation of blood flow is cold extremities. This doesn't typically affect us all the time, but many or most have episodic problems when we spend too much time upright. Doctors will pay attention if this reaches the point of causing loss of fingers or toes, but generally ignore every lesser problem because they are sure the patient's body will compensate, even if they have little idea how this works.

What I was suggesting in that previous post is that localized, episodic reduction in blood flow could tie together the problems in energy production, endocrine function and immune response. In the extreme case of stroke or other reperfusion injury a single episode can result in easily-detectable permanent damage. What happens in less severe cases of transient ischemias? The patients mostly recover without problems. How do they recover? That is a research question.

So, I'm saying "yes, chaos is involved in this disease" and "no, it is not unique to this illness". What I'm seeing is a defect in biological control of a number of important factors. I've concentrated on blood pressure and heart rate because these are easy to measure and understand. I could also talk about cortisol production or ADH. Lots of us have a lag in cortisol production which puts peak output well after the stimulus requiring it. Sum up production around a daily cycle, and we are usually normal. We can produce cortisol in response to demands, it just doesn't show up when needed, or doesn't stop when appropriate.

Likewise, with ADH (vasopressin), many of us are close to exhibiting polyuria because of inadequate ADH. This may even go so far as to result in diabetes insipidus. It appears there is a problem with regulation of total blood volume, but simply getting a test for such volume is outside of normal clinical practice. Regulation of electrolytes is also frequently disturbed. (In hot weather I take a "Salt Stick" capsule, along with fluids, before I go out. Most people don't have to be so careful.)

All of these things go beyond the simple "too much" or "too little" of standard clinical diagnostic tests. (It is as if your automobile mechanic only told you your car had an oil shortage which could be corrected by topping up the crankcase before each trip. Such an automobile needs repair, but I seriously doubt that mechanic should be the one to do it.)

Everywhere I look in this illness I see evidence of poorly regulated biological control systems. There is a great deal of interaction between these, as I've tried to argue about dynamics of blood flow. In a great many cases doctors can ignore dynamic behavior, secure in the knowledge the patient's body will somehow sort the problem out. This is not one of those cases. If you want to find a traditional cause, assuming there is such a single cause, you first need to see how these dynamics differ from health, and find what the abnormal behavior is responding to. If you wear blinders which prevent you seeing dynamics, you are not going to accomplish much.

 

[Dr.Patient]

"The truth is that we don't understand what is happening in patients who are sick, but likely to fully recover. Recovery can take place in hours, days, weeks or months, and we have little basis on which to predict which patients will take which times. The possibility that there are previously-healthypatients who will remain sick for years, without dying and providing convenient specimens for pathologists, is resolutely rejected. Thesepatients are anomalies to current medicaltheories (in the terminology of Thomas Kuhn) and are treated with the scorn typically reserved for anomalies you can't either predict or change."

Very well said! Picking up multisystem dysfunction patients, and following several parameters, over years and years, then finding out who recovers fully vs partially vs worsening---this is something that the medical system at this time is unable to handle. Getting some supercomputers to work on this is the only way.

 

[anciendaze]

Consider that patient who found she had severe POTS, using a home test -- after seeing both a PCP and a cardiologist. I don't think a supercomputer is required in that case.

She appears to have been given no useful advice on managing the condition, and was placed on a waiting list months long to see a specialist. She was not given advice about limiting time upright, or warnings about making bad decisions when suffering from hypoxia.

You would think Dr. Streeten had never done any work on these subjects.

Incidentally, if you come up with a useful computational task for what used to be a supercomputer, I have one with 8 cores, 32 GB RAM and 10 TB of rotating storage to dedicate to a task. It is hard to really keep these beasts usefully busy. I've calculated Pi to 5,000,000,000 places to test reliability, and in response to another suggestion I've solved the problem with a hand calculator and some thought. I'm still looking for uses for this machine I put together from used parts gamers now consider obsolete.