A test of the adaptive network explanation of functional disorders using a machine learning analysis of symptoms, 2018, Melidis, Hyland et al

A test of the adaptive network explanation of functional disorders using a machine learning analysis of symptoms

Author Information:
Melidis C, Denham SL, Hyland ME.
University of Plymouth

Abstract
The classification and etiology of functional disorders is controversial. Evidence supports both psychological and biological (disease) models that show, respectively, that functional disorders should be classified as one (bodily distress syndrome) and many (e.g., irritable bowel syndrome (IBS), fibromyalgia syndrome (FMS), and chronic fatigue syndrome (CFS)). Two network models (symptom network and adaptive network) can explain the specificity and covariation of symptomatology, but only the adaptive network model can explain the covariation of the somatic symptoms of functional disorders. The adaptive network model is based on the premise that a network of biological mechanisms has emergent properties and can exhibit adaptation. The purpose of this study was to test the predictions that symptom similarity increases with pathology and that network connection strengths vary with pathology, as this would be consistent with the notion that functional disorder pathology arises from network adaptation. We conducted a symptom internet survey followed by machine learning analysis. Participants were 1751 people reporting IBS, FMS or CFS diagnosis who completed a 61-item symptom questionnaire. Eleven symptom clusters were identified. Differences in symptom clusters between IBS, FMS and CFS groups decreased as overall symptom frequency increased. The strength of outgoing connections between clusters varied as a function of symptom frequency and single versus multiple diagnoses. The findings suggest that the pathology of functional disorders involves an increase in the activity and causal connections between several symptom causing mechanisms. The data provide support for the proposal that the body is capable of complex adaptation and that functional disorders result when rules that normally improve adaptation create maladaptive change.

 

If you have a checklist with symptoms, the symptom profiles will tend to converge as number of checked symptoms increases. Since symptom clusters is just a different way to represent this data, you're obviously going to find this correlation.

You cannot draw this conclusion without a control group of people with nonfunctional illness. Maybe what they're seeing is exactly the same that is seen in conditions like multiple sclerosis.

Plus the concept of functional disorder as illness category is stupid: it misleads people into thinking there must be some common features shared by all illnesses in this category, when in reality all it means currently unexplained illness.

 

I assume that this is psych-speak for "we've made up some conditions and this is the dubious reasoning for them". But equally, couldn't you say the same thing about a real/physical/insert appropriate description condition i.e. I have a cold at the moment. The pathology of my cold involves an increase in the activity and causal connections between several symptom causing mechanisms (i.e. the assorted ways my immune system reacts to the cold virus).

Oh, brainwave. Because it can be put that way, it theoretically adds weight to their argument that their made-up conditions exist, on the assumption that you accept that the symptom causing mechanisms that they claim to be able to identify actually exist.

 

A potential problem is people might not have all the diagnoses they should have. I was diagnosed first with IBS. I had ME/CFS at that stage but not the diagnosis. Also the definition of Fibromyalgia has widened with the 2010 criteria which don't require a tender point evaluation. A lot of doctors particularly it seems rheumatologists diagnose people with fibromyalgia even when they have CFS symptoms.

 

I doubt there is much sense to be made.

I suspect what they are proving is that diagnoses are made by people who use adaptive learning and sometimes go maladaptive. Its a bit like psychosomatic research I suppose - all in the heads of the researchers. Symptoms fall into certain clusters because the people who decide what out as symptoms (doctors) like to cluster them.

 

I pricked up my virtual ears at this paper, partly because I find systems stuff very interesting in general, and partly because the idea of dysregulation, whether immune, metabolic, endocrine/neurological and/or something else, being integral to ME as a potential prolonging factor of some sort is intriguing, not least because it might be fixable/reversible.

That is, some hit-and-run stressor like a virus, for example, knocks one or more of the body's set points away from normal, and for some reason the system stays off-kilter thereafter, even when the initial insult is gone. Other body systems domino-adapt to the new normal, causing the plethora of symptoms.

However, I am not a life scientist, so am prepared to stand corrected if this doesn't make sense (and if it does, I have no idea whether it is actually the case).

A bit of googling reveals that Michael Hyland, the third author on the paper, wrote a book on the subject in 2011 ('The Origins of Health and Disease), where CFS is one of his examples.

(paragraphs added)

In the book he says,

He does not appear to conflate CFS with MUS:

He goes on to say,

I've only read a little via the free kindle sample, so don't know what Hyland's detailed ideas about CFS actually are.

In the part of the intro available for free, which is well-written and interesting, he sounds pretty sane, eg, acknowledging that, in effect, bedside manner is responsible for 90% of the benefits of psychological therapies, and noting that scientists are not always open-minded:

 

PS @Valentijn, thought you'd particularly like this quote from the book,

ETA correction.

 

My first reaction to this was - so multiple overlapping symptoms make differential diagnosis difficult. Whoopee!

I think we're both being unfair.

One of the authors (last listed - I assume it's his baby) proposes that rather than being due to false illness beliefs, 'functional' illnesses can be explained by 'emergent' network interaction between biological systems.

I'd like to see some explanation of this explanation but I can at least see where he's coming from and it isn't from the BPS camp :

https://www.researchgate.net/profile/Michael_Hyland4

 

Very helpful @Marco, as he has a comment on that page in answer to a question:

 

Hyland is a psychologist who had moderate ME/CFS for 2 years, and both factors seem to be strongly influencing his beliefs about ME/CFS. While he has enough first-hand knowledge to reject the psychosomatic paradigm, he still sees the disease in a very psychosocial framework. The result seems to be a central sensitization theory - eg, all of the hardware is in good shape, but something in your brain or adrenal system ("network") is giving the wrong signals.

He also explicitly attributes the disease to lifestyle, probably based on his personal experience rather than any scientific evidence:

"Relearning how to listen to your body's signals" sounds an awful lot like learning to ignore symptoms and interpret them differently:

Fortunately the key to recovery is to emulate him and avoid stress, toxins, caffeine, alcohol, etc, and get exactly the right amount of exercise (despite being "very disinclined" due to the disease), because a car battery won't recharge itself if you don't drive it

 

And he thinks the accu will recharge when the dynamo does not work?

 

Maybe, I certainly agree I may have been a bit quick to respond. But on further deliberation I am not sure I end up with a very different view.

Highland says amongst other things:

This paper presents the idea that diseases can be explained and treated with biological theories, mental illness with psychological theories,

I do not buy that. Mental illness is not explained by psychological theories. Schizophrenia is not, severe depression is not, mania is not, obsessive compulsive disorder is not. None of them are explained at all, except perhaps by the finding of genetic predispositions and the use of psychotropic drugs like LSD. Explaining mental illness with psychological theories is the realm of Freud, which hopefully we have long since abandoned.

My take is that we take a biological systems dynamic approach for ALL illnesses. (And I guess maybe I think he is wrong about the causation of the 'diseases' too - they involve complex systems dynamics too, not just simple causes.)

But having worked with complex systems dynamics I have found words like 'emergence' completely unhelpful. Positive feedback loops are important and easy to understand. Co-operative interactions ditto, and so on. The problem with 'emergence' is that it really means 'not predicted' and that is no use to science because scientific theories are all about prediction and testing. So rather than saying there is emergence one needs to work out the dynamic rules. For immunology and for brains that can take years but I don't think you need any special expertise in artificial intelligence. You just need the sort of common sense that a vacuum cleaner designer might make use of.

 

I can't access the full paper but I pretty much agree on most points.

Where I might diverge is on the issue of 'emergent behaviour' in respect of complex system dynamics and our ability to understand them. On one hand I can accept that 'emergent' is a handy cop-out that avoids the mental effort required to understand and predict. On the other I struggle to accept that it's possible to fully understand to the point of being able to predict complex interactions between multiple systems involving feedback loops; adaption etc etc.

 

They're starting off with the unfounded assumption that there is a class of functional disorders at all.

I haven't seen any evidence supporting the psychological model of one broad syndrome, ever. Just some hacks playing at imagining grand theories to support their distorted perceptions and desired outcomes.

For anyone baffled by the terminology, just ignore the "network" bit - they're ultimately talking about symptoms being static things versus things which interact with each other. The body is an inherently adaptive system where many parts constantly interact and change as a result, so labeling symptoms of any disease as being an adaptive network is simply stating the obvious.

How do they establish increasing pathology? It's also making quite a leap to assume that network adaptation causes disease, rather than resulting from a disease. There's no apparent way to prove such a thing, so I very much doubt that they have. And since network adaptation is the default state of the human body, if it caused disease then disease would be the default state of being for everyone

A survey on the internet ... how scientific This means diagnosis was not a rigorous process and completely unverified, which is a pretty essential problem in research.

Symptom clusters are an artificial construct, and likely to be heavily influenced by the diagnostic criteria themselves. And if criteria are used which focus on vague symptoms, then of course there will be a lot of overlap. However, I doubt that more specific symptoms such as properly defined PEM, GI bleeding, or tender points (versus tender everywhere) have much overlap unless there is a valid co-diagnosis.

Again, they're stating the obvious - every disease causes symptoms, typically multiple symptoms which are accordingly related to each other by virtue of being caused by the same disease.

So instead of our brain malfunctioning by unreasonably thinking we're sick, they're saying that our body generates fake signals to make the reasonable brain think that we're sick. The end result is the same - you are misinterpreting your symptoms. The difference here from typical psychosocial bullshit is that there's some actual biology involved, in generating the false signals. But the solution is likely the same as given by Hyland previously - learn to reinterpret the signals, probably with some psychological help

 

My two pence worth is that people come up with these convoluted theories because as complex as they may appear to be, it is always much easier to come up with something like this which doesn't require any good evidence (you literately just make something up that sounds plausible,and likely convince yourself and some others that's it's right along the way as a result) than it is to do genuinely useful experiments which find something, because useful experiments usually find nothing, it's very hard to do good science which actually finds something useful, but it's still more useful to do a proper sound experiment that rules something out than this. This stuff here is much much easier. It's also almost certainly wrong, not just because there's a lack of evidence for it but because it's a unnecessarily complex and that means something, probably a lot, is going to be horribly wrong. What a waste of effort. My primary objection to this though is that there is no evidence, I'm willing to go wherever the evidence leads, but this research just takes us on an imaginary journey in the car, our eyes closed, imagining the scenery going by, but car isn't moving, the engine isn't even on. What's the point?

 

Nice points, @strategist!