Identifying Defining Aspects of Chronic Fatigue Syndrome via Unsupervised Machine Learning and Featu

[Bob]

I have the impression that this paper is intended to be exploratory, or interim, rather than a comprehensive and conclusive study. I get the feeling it is part of a larger, ongoing, project that Jason is involved with.

 

[WillowJ]

Which is why these studies comparing accuracy on CFS vs healthy strike me as missing the point - and yup, I think just about any doctor could beat 90% accuracy when they were presented with pre-diagnosed CFS patients vs healthy people (the choice in this study).

I agree totally with you that ME/CFS vs. healthy is not the most useful way to study the validity of a diagnostic criteria because in the real world, that's not the choice.

However, it's not been my experience that doctors and medical personnel are easily able to distinguish ME/CFS from healthy. I was told I was healthy by most everyone I saw for the first few years and I continue to have it was said or implied or be treated as if I were imagining things, exaggerating, or just plain making things up, and once it was even suggested I had somatoform disorder. This is an experience I hear from other patients.

As far as I can tell, this is not uncommon with any "invisible" disease, even "standard, accepted" ones with visible effects like RA.

It seems to happen more often to women and girls than to men (and boys? or are they at risk for their youth? "too young to be ill"?), however.

 

[Bob]

My doctor (general practitioner) never diagnosed me. He was clueless. After months of being ill, I told him that I had ME, and he just made a note on his PC. That was the extent of my diagnosis and discussions about ME with my doctor. I assume that he accepted my self-diagnosis, because he never challenged me. It would have helped him and me if he had been able to ask me five to eleven specific questions as a guide for a diagnosis.

 

[Bob]

There's a blog about this paper, posted by Erica Verrillo...

Definitions, Shmefinitions - Leonard Jason's Team Leaves Them All in the Dust
http://cfstreatment.blogspot.co.uk/2014/02/definitions-shmefinitions-leonard.html

 

[Firestormm]

There's a blog about this paper, posted by Erica Verrillo...

Definitions, Shmefinitions - Leonard Jason's Team Leaves Them All in the Dust
http://cfstreatment.blogspot.co.uk/2014/02/definitions-shmefinitions-leonard.html

Thanks Bob

Not sure I am as enthused as Erica but then to be fair to her I haven't even tried reading the paper and she presumably has. Still, it is interesting to note that she would seem to be suggesting (or my interpretation of her enthusiasm for this means of testing criteria is such that) the CCC is in need of review. Or at least methods such as those exemplified by this paper should be employed to determine a new (and better) definition. Perhaps once CDC has collected its data they will use such means? Or IOM will consider Lenny's work here in helping them better determine applicability... I just don't buy into it myself for the reasons outlined above, namely that we are trying to be too clever with subjective questions and that the whole debate will change once objective biomarkers are better determined and applied.

N.B. Did you notice in the ranking of questions to results, the top 15 on that table, that PEM e.g. 'delayed reaction to activity causing symptom exacerbation', was not a feature?

At least not recognisably so. I guess we all have our own way of defining PEM, but there was no 24-hour delay etc. in the results. Only next day soreness from doing everyday activities.

No PEM 'caused by over-exertion with reaction delayed by 24-48 hours' etc. etc. And yet I thought this paper was trying to be more indicative or reflective of e.g. CCC and ICC who both hold PEM/PENE to be key features of 'the' condition.

Perhaps it is not as key as some of us might like to think or at least not as prominent on Lenny's Questionnaire at any rate?

 

[Bob]

@Firestormm, yes, I think this paper raises more questions than it answers, and I'm not as enthusiastic about it as Erica either, but I do find it very interesting.

I think some answers might lie within Jason's previous papers related to the DePaul Symptom Questionnaire (DSQ), but I've not read up on the DSQ. I'm not sure if I've had access to the relevant papers, despite looking for them occasionally.

I think that question 2 ("Next day soreness or fatigue after non-strenuous, everyday activities") could perhaps be considered to be related to a delayed post-exertional reaction, but I agree with you that the wording isn't specific enough to define a delayed reaction.

Perhaps it isn't helpful to specify a period of delay in a clinical criteria because the period of delay is different for everyone, and it would confused both patients and clinicians? And I think it's very hard for new patients, for example, to be aware of any delay because they don't yet associate today's flare-up with yesterday's exertion.

I'm not sure if this paper is trying to be reflective of the CCC and ICC, exclusively. They seem to be including Fukuda patients. I still don't understand how Jason decided who had ME/CFS for the purposes of this study or for the DSQ in general.

 

[Firestormm]

@Firestormm, yes, I think this paper raises more questions than it answers, and I'm not as enthusiastic about it as Erica either, but I do find it very interesting.

I think some answers might lie within Jason's previous papers related to the DePaul Symptom Questionnaire (DSQ), but I've not read up on the DSQ. I'm not sure if I've had access to the relevant papers, despite looking for them occasionally.

I think that question 2 ("Next day soreness or fatigue after non-strenuous, everyday activities") could perhaps be considered to be related to a delayed post-exertional reaction, but I agree with you that the wording isn't specific enough. Perhaps it isn't helpful to specify a period of delay in a clinical criteria because the period of delay is different for everyone, and it would confused both patients and clinicians? And I think it's very hard for new patients, for example, to be aware of any delay because they don't yet associate today's flare-up with yesterday's exertion.

I'm not sure if this paper is trying to be reflective of the CCC and ICC, exclusively. They seem to be including Fukuda patients. I still don't understand how Jason decided who had ME/CFS for the purposes of this study or for the DSQ in general.

I must confess - well not to you because you already know I think - that I really only pay cursory attention to all these kind of developments. So I also haven't read or even really grasped what the DSQ is or how useful it might be - I don't know even if the CDC have embraced it or if Jason is still trying to sell it as something of benefit. I see that folk are criticising (unfairly in my view) the CDC for 'going back to basics' and yet this is also what Jason seems to be doing. Rightly or wrongly, the number of people diagnosed now with 'CFS' is vastly more than might once have been had Ramsay been more successful with his initial research and definition (that's largely my opinion - take from it what you will).

We can't turn back time, and I think there is merit in completing these kind of assessments now, and in bringing all this investment into the field properly (yes - I do think HHS, IOM and CDC efforts are 'good' generally and whilst my fingers are crossed - I do think the outcome will be beneficial), I also mourn the early days.

If only... If only the research had supported the conjecture... It makes me very sad, but I haven't lost hope. There is something there and it might all seem unnecessary to some who understandably want to see rapid progress and answers, but it's very necessary to establish the foundations. I don't think it has been done before: hell we still have no solid epidemiology!

 

[aimossy]

In my humble view the epidemiology is what Unger seems to be doing or beginning to collect, and foundations are crucial in building pictures of disease. Someone has to get this part started, who else can do that apart from CDC? Which other places in the world are supposed to do this?....I just wish it happened a LOT earlier than this.

 

[Sparrow]

I think an additional challenge in crafting good diagnostic questions is that the symptoms have to be described in a way that brand new patients can recognize them.

I know that personally, for some symptoms, it took me a while to realize what was what. For example, for a while my next day PEM tended to feel mostly like I was sick with a virus of some kind (weak, exhausted, swollen glands, viral feeling, etc.), so for a long time I just thought that I was getting sick all the time. I didn't realize that was what PEM could feel like, and it took me a while to make the connection that it happened after a day of more activity every time. I didn't catch the cause and effect pattern of many of my symptoms right away, even though in hindsight it was obvious. I just thought that I had ups and downs where the symptoms were better or worse.

Similarly, I consistently told my doctor that I didn't have headaches. It took me forever to realize that the horrible sensation I had dubbed "brainfire" was probably what they meant by "headaches of a new type or severity". Because it didn't feel similar like a similar sensation to any previous headache, I just didn't make the connection. Again, it seems obvious now, but at the time it was such a distinct and different thing that I never thought to label it that way.

Edit: In some ways the kinds of questions this team was suggesting look like a step forward on that side (though I suspect they could be improved). It would be great if the diagnosis was more certain and more natural, and if it was easy for patients to see their symptoms reflected there. In some ways, I wish we could just say "legs feel like they have lead weights attached" or "feel like gravity is way too strong" or "feel like I'm about to get a cold" or "can't stay awake no matter how hard I try" or "muscles feel like they're powered by dying rechargeable batteries" and put things like that in a box to choose from in place of "fatigue". If we could find the right group of descriptors, I bet we'd get a lot less false positives that way.

 

[MeSci]

I think an additional challenge in crafting good diagnostic questions is that the symptoms have to be described in a way that brand new patients can recognize them.

I know that personally, for some symptoms, it took me a while to realize what was what. For example, for a while my next day PEM tended to feel mostly like I was sick with a virus of some kind (weak, exhausted, swollen glands, viral feeling, etc.), so for a long time I just thought that I was getting sick all the time. I didn't realize that was what PEM could feel like, and it took me a while to make the connection that it happened after a day of more activity every time. I didn't catch the cause and effect pattern of many of my symptoms right away, even though in hindsight it was obvious. I just thought that I had ups and downs where the symptoms were better or worse.

I think it was years after I became ill that I even heard of PEM, and more years (requiring keeping a more detailed health diary) before I realised that my malaise was due to exertion 1-3 days earlier.

 

[Sparrow]

I think it was years after I became ill that I even heard of PEM, and more years (requiring keeping a more detailed health diary) before I realised that my malaise was due to exertion 1-3 days earlier.

This just made me realize something I hadn't considered before about using PEM as a required symptom.

I suspect what you're describing is really common. It's not an obvious cause and effect at first, and the delay makes it even harder to notice. I think PEM really is a hallmark symptom, but I wonder if our attempts to make sure it's emphasized in diagnosis might cause problems that way. I wonder how many not-yet-diagnosed patients will actually recognize it well enough to be able to tell their doctors that's what they're experiencing.

Even after I knew what PEM was in theory, and had learned to easily recognize what really bad PEM looks like for me, it still took me years to realize that other activities I was doing day to day were responsible for a pile of my daily symptoms. Because I was in a constant state of low-grade PEM, I didn't realize those symptoms weren't just a constant part of my illness. Until I cut back even further and a pile of symptoms cleared up, I had no idea that was even possible. I wish I had realized sooner. My life could probably have been a lot less hellish for those first couple of years.

 

[Firestormm]

I think an additional challenge in crafting good diagnostic questions is that the symptoms have to be described in a way that brand new patients can recognize them.

I know that personally, for some symptoms, it took me a while to realize what was what. For example, for a while my next day PEM tended to feel mostly like I was sick with a virus of some kind (weak, exhausted, swollen glands, viral feeling, etc.), so for a long time I just thought that I was getting sick all the time. I didn't realize that was what PEM could feel like, and it took me a while to make the connection that it happened after a day of more activity every time. I didn't catch the cause and effect pattern of many of my symptoms right away, even though in hindsight it was obvious. I just thought that I had ups and downs where the symptoms were better or worse.

Similarly, I consistently told my doctor that I didn't have headaches. It took me forever to realize that the horrible sensation I had dubbed "brainfire" was probably what they meant by "headaches of a new type or severity". Because it didn't feel similar like a similar sensation to any previous headache, I just didn't make the connection. Again, it seems obvious now, but at the time it was such a distinct and different thing that I never thought to label it that way.

Edit: In some ways the kinds of questions this team was suggesting look like a step forward on that side (though I suspect they could be improved). It would be great if the diagnosis was more certain and more natural, and if it was easy for patients to see their symptoms reflected there. In some ways, I wish we could just say "legs feel like they have lead weights attached" or "feel like gravity is way too strong" or "feel like I'm about to get a cold" or "can't stay awake no matter how hard I try" or "muscles feel like they're powered by dying rechargeable batteries" and put things like that in a box to choose from in place of "fatigue". If we could find the right group of descriptors, I bet we'd get a lot less false positives that way.

I agree completely. You see for me, every day feels like flu without the snot. But PEM is different because of the intensity with which my symptoms flare and because it occurs more than a day after overdoing something. So when people talk about PEM being THE distinguishing factor of ME compared to CFS and how it can be measured using CPET I remain sceptical. But at some point a decision will need to be reached.

We do all have our own interpretation of what PEM means. For example, sometimes a relapse may not be seemingly caused by overexertion but by an external factor - and yet the 'flare' in symptoms can be similar - though perhaps more intense again - than the 'normal' PEM. Is it then exertion that causes these 'flares' and relapses or something else? Have you ever overexerted and not experienced PEM? How can people improve the amount they can do and for the period of time that they can do it - if the risks remain the same? Because it happens. I am better functioning now than I was 4 years ago when I last relapsed. What might be the biological reason for this?

And who knows what the effect of sleep disruption might be having on headaches/morning hang-overs, daily fatigue, cognition, and PEM? There is much about our illness let alone PEM than we don't yet understand enough to be able to reasonable compare with others and say 'mine is the same as yours therefore we have X and they have Y'. Which is why I think this 'back to basics' work is necessary - but only if the biomedical research is also ramped up to try and also better understand the processes that might be involved and how they interact to produce the symptoms.

 

[Dolphin]

I believe that this study was based on dividing up being patients and controls based on the (relative) severity and frequency of the symptoms. I think this means that somebody with the illness had to have every symptom when they were coming up with the list of symptoms. However, it might be the case that one could get higher sensitivities and specificities if one didn't require the presence (relatively high severity and frequency) of some symptoms. For example, as was mentioned, some people might not recognise themselves as having a symptom (I had some similar experiences myself). Or indeed, they might not have it but it can be a useful symptom in those that do have it (e.g. not everyone with diabetes has retinopathy or neuropathy but checking for such problems could be useful in diagnosing diabetes if one didn't have a blood test).

This sort of lack of requirement for specific symptoms that are still utilised is seen in the ICC and Canadian criteria* where there are groups of symptoms and somebody only has to have one or two from the category. It is disappointing they didn't investigate whether they could have improved on this using this, but perhaps the technique is not used easily in this way.

By the way, the two main authors are undergraduates.

*it is also seen in the Fukuda but not requiring a post-exertional symptom means these criteria have a fundamental problem in my view.

 

[Dolphin]

It's a pity that the authors don't give what were the optimum severity and frequency thresholds for each individual symptom. Without it, it's hard to know how others can optimally use the results.

 

[leela]

I think this is HUGE.
Here is a blog about it I just read:
http://cfstreatment.blogspot.com/2014/02/definitions-shmefinitions-leonard.html

90% accuracy is amazing, relative to what we've had so far, and relative to most methodologies; especially considering what has not yet been
*added* to the data in the DSQ.

 

[Firestormm]

@leela see also posts #24 and #25 above on this thread

 

[Bob]

This seems to be the DePaul Symptom Questionnaire:
http://condor.depaul.edu/ljason/cfs/measures.html

Questions 72 to 76a seem to deal with PEM.

 

[Bob]

Here's a similar recent paper that @Simon has analysed:
http://forums.phoenixrising.me/inde...finition-for-cfs-me-leonard-jason-2013.27006/

It has slightly different conclusions:

Employing these higher frequency/severity requirements, the CART algorithm identified three symptoms that accurately classified 95.4% of participants as patient or control: fatigue/extreme tiredness, inability to focus on multiple things simultaneously, and experiencing a dead/heavy feeling after starting to exercise.

Snippet from Simon's analysis of Jason's Dec 2013 paper:

CART model predicts CFS using three symptoms
Using a CART, and counting symptoms only when they were at least moderate and present for at least half the time, just 3 symptoms (out of 54) were able to predict if a person was a CFS case or a heathy control:
- fatigue or extreme tiredness
- inability to focus on more than one thing at a time [neurocognitive problem]
- experiencing a dead or heavy feeling after starting to exercise ["Post-exertional Malaise" problem]

 

[Firestormm]

@Bob I was reading briefly an article this morning that inferred according to the CCC, PEM was not defined as including any delay to onset, but it was recovery that was delayed. I don't have time at the moment to check that, but wondered if you knew and happened to be around? Thanks

 

[Bob]

@Bob I was reading briefly an article this morning that inferred according to the CCC, PEM was not defined as including any delay to onset, but it was recovery that was delayed. I don't have time at the moment to check that, but wondered if you knew and happened to be around? Thanks

Yes, the CCC says that fatigue after exertion can be immediate, with reference to PEM:
"Fatigue immediately following activity may also be associated with these signs of immune activation."

And the CCC specifically mentions a slow recovery after a post-exertional flare-up, with reference to PEM:
"There is a pathologically slow recovery period - usually 24 hours or longer."