Brainstorming

[Cort]

If this project gets done what kind of variables should it contain? What kind of questions should it be able to answer?

Perhaps the best way is to ask yourself what do you want to know about ME'/CFS and the community and go from there.

Here are some ideas from the first page

CFS Itself Need to Be Characterized Better/Subsets need to be revealed - PLM does not attempt to characterize some important community variables or to pluck out Subsets - CFS, as we all know is a very complex disorder, possibly with many subsets. Creating a site specifically attuned to CFS could help us better understand that aspect of it.

PLM asks what kinds of treatment work for people with fatigue in CFS. We would ask that as well but we more moving from a primarily Treatment program to a disease characterization program: We could ask simple questions such as if type of onset effects the disease progression at all? How about Illness duration? How are people whove had this disease 5 years different from people whove had it 20 years. Dr. Maes says rates of IBS increase - do they? Ill bet rates of MCS and FM increase. Do they?

Examples - Gender - a very simple example, create a table or pictorial representation examining the effects of gender: males vs female on various factors . (Unfortunately I can't drop my table on here. Imagine it as a table.

Two columns with the stats underneath Male l Female
Average age
Average duration of illness
Average activity level
Type of Onset %acute w/out warning, %acute w/ previous reduced health, %gradual %acute w/out warning, %acute w/ previous reduced health, %gradual
Symptom presentation Pie chart showing relative proportions of immune/nervous system, etc symptoms (CCC)?
Or a list of most common symptoms reported
Co-occurring Conditions
?????

Query specific sets of patients - You could look for all female patients say with active HHV-6 infection and ill for over 10 years

And ask the database to list the treatments theyve tried, what their average effectiveness is and something about the patients, say what their average activity level is, what their most prominent symptoms, etc., what percentage of these patients are able to work, their co-occurring conditions, etc.

Directly compare different type of CFS patients; we could bring up

Patients with HHV-6 and EBV infection and then remove the EBV infection variable from the equation and see get some idea of what EBVs effects are; do certain symptoms recede? Do activity levels go up?

Better characterize the positive responders - scan the system for people who have substantial jumps in activity levels (2 or more) during a treatment. Provide a list in descending order of highly effective treatments in the CFS community. Characterize the entire group of people who experience this? %acute/gradual onset, top 10 moderate to severe symptoms, average duration of illness, percent male/female,

Knock it Out of the Park Section - Provide a list of treatments and the number of people who have experienced very high amounts of success on them (who have improved 4 or 5 activity levels on them). Characterize the entire group of people who experience this? %acute/gradual onset, top 10 moderate to severe symptoms, average duration of illness, percent male/female,
Compare to the CFS Community at large - similar list for Community at large
Best Bang for the Buck Section - figure in cost and benefit to determine the cheapest ways to see improvement

Characterize the Poor Responders -

Characterize, characterize, characterize - ideas

Characterize People Who Report They have Gotten Worse over time -what are they like?
Characterize People Who Report they have improved over time - how are they different?
Characterize People with FM vs without, with MCS vs without, with IBS vs without.
Characterize People who have the illness for < 1 year, 1-5 years, 5-10 years, 10-20 years, >20 years
Look at everything: activity levels, symptoms, lab tests, disagnoses cropping up, even the participation in the work force - we get a snap shot of what happens to ME/CFS patients simply by focusing on duration of illness and then querying the database.

Treatment Watch List - Create a Watch List or something - of treatments that commonly produce higher side effects.

 

[JanisB]

I just want to add that we should also look at relapses and crashes.

• What brings them on?
• How long do they last?
• How far down (e

.g. how many notches or % points down) do people fall?

I bring this up because I've gotten to 90% three times in the 22 years I've had this illness but I've relapsed four times. Each time different strategies seem to work. I think that if we could understand what triggers relapses and crashes, we could get a better handle on the pathophysiology of the illness.

I would also like to see what kinds of patterns emerge with functional medicine test panels such as urinary amino acids, plasma amino acids, organic acids (also called metabolic analysis profile), and comprehensive digestive analysis. A lot of research papers have been done on salivary and serum hormone, especially in the HPA axis, but nothing has been coordinated for the other kinds of testing typically done by holistic practitioners, and which have become common in autism, another NEID. It would be of great value to see if ME-CFS patients have similar or different patterns than ASD kids.

JanisB

 

[Mithriel]

Patients with HHV-6 and EBV infection and then remove the EBV infection variable from the equation and see get some idea of what EBVs effects are; do certain symptoms recede? Do activity levels go up?

We need to include enterovirus as well, especially Coxsackie B, since there is so much evidence that it is involved in classic epidemic/Ramsay ME.

Those in the US have to remember that this illness was being researched in the UK for decades before CFS.

Mithriel

 

[Esther12]

This sounds like such a mammoth task - heroic to take it on, slightly tragic we seem to need to.

Could we break it down, and make it flexible enough to change it as we go? - So we can add new characteristics and symptoms (or sub-characteristics and symptoms if we just start adding more detail?) without messing it all up. Doing something like that is easy with excel, but I've no idea how hard it would be to integrate it into a website.

I'll start listing characteristics for patients (a lot of people seem to change over time - should we have 'ever had' or 'currently have'? Or options for people to clarify? Some of them could pre-date CFS - should we clarify? A lot will have varying levels between patients). I've gone through about five or six different lists of possible symptoms, compiling what I could find. I'm sure it's far from exhaustive - but how thorough do we want to be?

Sudden or gradual onset?
Glandular fever?
Other viruses? All sorts. How stringent should we be on diagnosis? - should we have sub-sections for people positive with different tests? I've got the impression some docs are claiming people are positive for bugs they may not have (Lyme?)
Post-extertional malaise?
Overactive Thyroid?
Underactive Thyroid?
Troubled sleep? (Could add lots of sub categories here).
'Wired' tired?
Serious muscle pain (not just 'tired aches')?
Migraines? (Room for sub-sections here too)
Headaches (Ask for % of the time? Rate severity?)
Cognitive difficulties? (Sub-sections - Memory problems? Concentration? etc)
Painful glands (ask where? How often? How severe?)
Sore throat? (swollen? How often? How severe?)
Nausea? (Actual vomiting? How often?)
Funny body temperature stuff? (I'm currently wearing 7 tops - How many duvets do you use?)
Sweating?
Muscle twitching?
IBS type stuff (lots of sub categories possible here)
Depression? (History of, or only following diagnosis of CFS?)
Anxiety? (History of, or only following diagnosis of CFS?)
Palpitations?
Flu like fatigue?
Allergies?
Ear aches?
Dry eyes?
Dry mouth?
Sensitivity to light? (1-10?)
Sensitivity to sound (1-10?)?
Skin sensitivity/irritability?
Visual disturbances?
Hyper mobility? Other connective tissue disorders?
POTS?
Alcohol intolerance?
Jaw pain?
Tinnitus?
Weight loss?
Weight gain?
Excessive hunger?
Often seem to catch other bugs?
Rarely seem to catch other bugs?
Gotten Worse over time?
Improved over time?
Length of illness?
Age at getting ill?


I'd be interested to have some social data too, and see if that affects anything:
In a relationship?
Wealth levels?
Supported by friends/ family?
Supported by doctor?
Children?
Social security (struggling to get benefits? Insecurity with your property, etc)?
Able to work? Supported by employer?
Activity levels? - Regulated activity? Pacing? GET? Doing what you feel like?


I think it would be good to find a computery person who'd be able to guide us on the best way to go about something like this before we do too much planning. It could be there are real limitations on what can be easily done, of it could be easy to come up with a highly adaptable tool for data collection and analysis.

Edit - feeling a bit dim at the moment, and not really sure if any of the above was remotely useful or relevant. I think I'll step away from the PC for a bit!

 

[Cort]

Perfect Esther - you added some things in there that I hadn't thought of. I'm getting a Wiki set up on the Forum that we can use to work on this stuff more easily. thanks!

 

[Suella]

I would be extremely interested in the work of Dr. Jonathan Kerr, where he discovered 7 different CFSME genotypes. The descriptions of these 7 different "types" of CFS/ME looked quite vague in the first reporting. I suspect I know which categoy I fit into but am not absolutely sure.

It may be that better knowledge of what makes up these 7 different groups will inform us in our attempt to characterize the illness further.

Suella

 

[lono]

Subsets by biomarkers

I'd be very interested in trying to create some CFS/ME subsets based upon abnormal biomarkers. As we are all well aware, this is what gets respect in the medical community. However, decades of medical research studies have shown that CFSers don't necessarily share a common set of abnormal biomarkers.

I strongly believe this is because we share similar "end conditions": horrible fatigue, PEM, etc., but there are many reasons to believe that not all of us arrived at this end condition via the same mechanisms. Many of use have important biomarker abnormalities, we just don't all have the same ones.

I know there are many (but definitely not all) CFSers who, like me, have abnormalities in most of these biomarkers:

- low vasoactive intestinal peptide (VIP)
- low melanocyte stimulating hormone (MSH)
- elevated C4a
- elevated TGF Beta 1

I bet there are other subsets of people on this board who have constellations of different, abnormal biomarkers.

And one other biomarker item I am very interested in-- multiple chemical sensitivity is in an even worse position re: having no recognized biomakers. I keep hearing a fair amount of anecdotal evidence that MCS people have low vasoactive intestinal peptide (VIP) levels. I know that is true in my case, but does that match with other MCS folks on this board? Note that not everyone with low VIP seems to have MCS (you're lucky!!!), but I keep hearing that people with MCS seem to test abnormally low in vasoactive intestinal peptide.

 

[Sam Carter]

I wonder if neural networks could help us? A neural network is a computer program which looks for patterns and correlations in data. I think they could be really useful at picking up the subtle but clinically important characteristics of ME patients that a human observer would not see. Two papers of interest are:

Detection of immunologically significant factors for chronic fatigue syndrome using neural-network classifiers.
Clin Diagn Lab Immunol. 2001 May;8(3):658-62.
Hanson SJ, Gause W, Natelson B.

Generation of classification criteria for chronic fatigue syndrome using an artificial neural network and traditional criteria set.
In Vivo. 2002 Jan-Feb;16(1):37-43.
Linder R, Dinser R, Wagner M, Krueger GR, Hoffmann A.

Lenny Jason wrote:

""""""""""""""""""""
Linder et al. (2002) used artificial neural networks
to classify patients with chronic fatigue (including CFS
and idiopathic chronic fatigue), lupus erythematosus, and
FM. They were able to achieve a sensitivity of 95% and a
specificity of 85%. Those chronic fatigue symptoms that
had the highest accuracy were “acute onset of symptoms”
and “sore throat,” which supports the hypothesis of an
infectious etiology.
""""""""""""""""""""
from: Chronic Fatigue Syndrome: The Need for Subtypes
Neuropsychology Review, Vol. 15, No. 1, March 2005
Leonard A. Jason, Karina Corradi, Susan Torres-Harding,
Renee R. Taylor, and Caroline King

 

[Sam Carter]

Btw, I meant to say that I think a wiki is a fantastic idea! There is so much useful info on the forum that gets hidden away amongst the threads - consolidating it into a single place would be amazing.

 

[Advocate]

I wonder if neural networks could help us? A neural network is a computer program which looks for patterns and correlations in data. I think they could be really useful at picking up the subtle but clinically important characteristics of ME patients that a human observer would not see.

Hi Sam

Interesting. Have you had any personal experience with artificial neural networks?

Advocate

 

[Sam Carter]

Hi Sam

Interesting. Have you had any personal experience with artificial neural networks?

Advocate

Hello Advocate,

I don't, alas, have any personal or practical experience with ANNs. They've intrigued me for a while because I think if they were given a sufficiently large dataset they could delineate a characteristic, and therefore diagnostic, biomedical signature.

Sam

 

[Cort]

I wonder if neural networks could help us? A neural network is a computer program which looks for patterns and correlations in data. I think they could be really useful at picking up the subtle but clinically important characteristics of ME patients that a human observer would not see. Two papers of interest are:

Detection of immunologically significant factors for chronic fatigue syndrome using neural-network classifiers.
Clin Diagn Lab Immunol. 2001 May;8(3):658-62.
Hanson SJ, Gause W, Natelson B.

Generation of classification criteria for chronic fatigue syndrome using an artificial neural network and traditional criteria set.
In Vivo. 2002 Jan-Feb;16(1):37-43.
Linder R, Dinser R, Wagner M, Krueger GR, Hoffmann A.

Lenny Jason wrote:

""""""""""""""""""""
Linder et al. (2002) used artificial neural networks
to classify patients with chronic fatigue (including CFS
and idiopathic chronic fatigue), lupus erythematosus, and
FM. They were able to achieve a sensitivity of 95% and a
specificity of 85%. Those chronic fatigue symptoms that
had the highest accuracy were “acute onset of symptoms”
and “sore throat,” which supports the hypothesis of an
infectious etiology.
""""""""""""""""""""
from: Chronic Fatigue Syndrome: The Need for Subtypes
Neuropsychology Review, Vol. 15, No. 1, March 2005
Leonard A. Jason, Karina Corradi, Susan Torres-Harding,
Renee R. Taylor, and Caroline King

Great idea Sam, I hope we can get some statisticians to look for these patterns. We do have some good news on this front - I've put the proposal before several researchers they all think the program can generate useful data for researchers; we'll be working with them I believe, to create the system.

 

[Cort]

Btw, I meant to say that I think a wiki is a fantastic idea! There is so much useful info on the forum that gets hidden away amongst the threads - consolidating it into a single place would be amazing.

I purchased a wiki designed for the Forums about a month ago; we've just had problems getting it installed (very frustrating!) but it will get done.

 

[Zig]

I've managed a pattern recognition project that included ANN before. They can be really useful for digging up patterns that are very hard to see using classical techniques. In fact, that's what we used it for. If we couldn't find results using classical and specialized techniques, we switched on the ANN engine.

The good thing is that ANN engines can be quite adaptable and can be added in at the end, if desired. You probably won't have to change any of the input parameters (symptoms list, treatments list, biomarkers list, etc). Whereas, for the classical and specialized techniques, you kind of want to have at least the pertinent input parameters.

Okay, I am going to step back for a second, and take a look at the big picture, here. What I or anyone wants from this treatment program is something that works for each of us individually. Whether this is the same treatment for all of us, or only a small cluster of us, doesn't matter as long as it works for us individually. At first, we'll probably find treatments that help small clusters of us. In time, as we add different treatments, we'll see the bigger clusters. We're kind of mining the CFS disease from the bottom up, a grassroots approach, if you wish.

At the top end, we have the WPI's of the world mining from the top down. They use scientific theories to uncover new biomarkers (eg. XMRV) and treatments that may treat all of CFS successfully. There's no guarantee they'll crack the mine from the top, but thats Ok because we are making steady progress from the bottom up.

I didn't locate the thread, yet, about what are our best treatments, but sifting through that, one might be able to figure out what some of the important input parameters are.

I haven't looked at PLM to see how they organize their data, but I should.

 

[Cort]

Sorry not to get back to this earlier but I hope you will look at PLM and see what you think.

 

[Leopardtail]

If this project gets done what kind of variables should it contain? What kind of questions should it be able to answer?

Perhaps the best way is to ask yourself what do you want to know about ME'/CFS and the community and go from there.

Co-occurring Conditions

Primary (most relevant diagnosis): E.g. Fibro, ME (with PEM) or CFS (without PEM)
And co-occuring conditions that are not ME/FMS symtpoms (e.g. diabetes) are a must for me.

Query specific sets of patients - You could look for all female patients say with active HHV-6 infection and ill for over 10 years

And ask the database to list the treatments theyve tried, what their average effectiveness is and something about the patients, say what their average activity level is, what their most prominent symptoms, etc., what percentage of these patients are able to work, their co-occurring conditions, etc.

Directly compare different type of CFS patients;

Great idea

Better characterize the positive responders ......
Knock it Out of the Park Section - Provide a list of treatments and the number of people who have experienced very high amounts of success on them (who have improved 4 or 5 activity levels on them). Characterize the entire group of people who experience this? %acute/gradual onset, top 10 moderate to severe symptoms, average duration of illness, percent male/female,
Compare to the CFS Community at large - similar list for Community at large
Best Bang for the Buck Section - figure in cost and benefit to determine the cheapest ways to see improvement

Characterize the Poor Responders -

I we are going to do this, we need to record other treatments that person was using at the same time. It must be possible to distinguish interaction with symptoms vs interaction with other drugs.

 

[Leopardtail]

I've managed a pattern recognition project that included ANN before. They can be really useful for digging up patterns that are very hard to see using classical techniques. In fact, that's what we used it for. If we couldn't find results using classical and specialized techniques, we switched on the ANN engine.

The good thing is that ANN engines can be quite adaptable and can be added in at the end, if desired. You probably won't have to change any of the input parameters (symptoms list, treatments list, biomarkers list, etc). Whereas, for the classical and specialized techniques, you kind of want to have at least the pertinent input parameters.

Okay, I am going to step back for a second, and take a look at the big picture, here. What I or anyone wants from this treatment program is something that works for each of us individually. Whether this is the same treatment for all of us, or only a small cluster of us, doesn't matter as long as it works for us individually. At first, we'll probably find treatments that help small clusters of us. In time, as we add different treatments, we'll see the bigger clusters. We're kind of mining the CFS disease from the bottom up, a grassroots approach, if you wish.

At the top end, we have the WPI's of the world mining from the top down. They use scientific theories to uncover new biomarkers (eg. XMRV) and treatments that may treat all of CFS successfully. There's no guarantee they'll crack the mine from the top, but thats Ok because we are making steady progress from the bottom up.

I didn't locate the thread, yet, about what are our best treatments, but sifting through that, one might be able to figure out what some of the important input parameters are.

I haven't looked at PLM to see how they organize their data, but I should.

Obvious point here, which exact symptoms did it help and/or exacerbate?

 

[Leopardtail]

Subsets by biomarkers

I'd be very interested in trying to create some CFS/ME subsets based upon abnormal biomarkers. As we are all well aware, this is what gets respect in the medical community. However, decades of medical research studies have shown that CFSers don't necessarily share a common set of abnormal biomarkers.

I strongly believe this is because we share similar "end conditions": horrible fatigue, PEM, etc., but there are many reasons to believe that not all of us arrived at this end condition via the same mechanisms. Many of use have important biomarker abnormalities, we just don't all have the same ones.

I know there are many (but definitely not all) CFSers who, like me, have abnormalities in most of these biomarkers:

- low vasoactive intestinal peptide (VIP)
- low melanocyte stimulating hormone (MSH)
- elevated C4a
- elevated TGF Beta 1

I bet there are other subsets of people on this board who have constellations of different, abnormal biomarkers.

And one other biomarker item I am very interested in-- multiple chemical sensitivity is in an even worse position re: having no recognized biomakers. I keep hearing a fair amount of anecdotal evidence that MCS people have low vasoactive intestinal peptide (VIP) levels. I know that is true in my case, but does that match with other MCS folks on this board? Note that not everyone with low VIP seems to have MCS (you're lucky!!!), but I keep hearing that people with MCS seem to test abnormally low in vasoactive intestinal peptide.

This sounds like such a mammoth task - heroic to take it on, slightly tragic we seem to need to.

Could we break it down, and make it flexible enough to change it as we go? - So we can add new characteristics and symptoms (or sub-characteristics and symptoms if we just start adding more detail?) without messing it all up. Doing something like that is easy with excel, but I've no idea how hard it would be to integrate it into a website.

I'll start listing characteristics for patients (a lot of people seem to change over time - should we have 'ever had' or 'currently have'? Or options for people to clarify? Some of them could pre-date CFS - should we clarify? A lot will have varying levels between patients). I've gone through about five or six different lists of possible symptoms, compiling what I could find. I'm sure it's far from exhaustive - but how thorough do we want to be?

Sudden or gradual onset?
Glandular fever?
Other viruses? All sorts. How stringent should we be on diagnosis? - should we have sub-sections for people positive with different tests? I've got the impression some docs are claiming people are positive for bugs they may not have (Lyme?)
Post-extertional malaise?
Overactive Thyroid?
Underactive Thyroid?
Troubled sleep? (Could add lots of sub categories here).
'Wired' tired?
Serious muscle pain (not just 'tired aches')?
Migraines? (Room for sub-sections here too)
Headaches (Ask for % of the time? Rate severity?)
Cognitive difficulties? (Sub-sections - Memory problems? Concentration? etc)
Painful glands (ask where? How often? How severe?)
Sore throat? (swollen? How often? How severe?)
Nausea? (Actual vomiting? How often?)
Funny body temperature stuff? (I'm currently wearing 7 tops - How many duvets do you use?)
Sweating?
Muscle twitching?
IBS type stuff (lots of sub categories possible here)
Depression? (History of, or only following diagnosis of CFS?)
Anxiety? (History of, or only following diagnosis of CFS?)
Palpitations?
Flu like fatigue?
Allergies?
Ear aches?
Dry eyes?
Dry mouth?
Sensitivity to light? (1-10?)
Sensitivity to sound (1-10?)?
Skin sensitivity/irritability?
Visual disturbances?
Hyper mobility? Other connective tissue disorders?
POTS?
Alcohol intolerance?
Jaw pain?
Tinnitus?
Weight loss?
Weight gain?
Excessive hunger?
Often seem to catch other bugs?
Rarely seem to catch other bugs?
Gotten Worse over time?
Improved over time?
Length of illness?
Age at getting ill?


I'd be interested to have some social data too, and see if that affects anything:
In a relationship?
Wealth levels?
Supported by friends/ family?
Supported by doctor?
Children?
Social security (struggling to get benefits? Insecurity with your property, etc)?
Able to work? Supported by employer?
Activity levels? - Regulated activity? Pacing? GET? Doing what you feel like?


I think it would be good to find a computery person who'd be able to guide us on the best way to go about something like this before we do too much planning. It could be there are real limitations on what can be easily done, of it could be easy to come up with a highly adaptable tool for data collection and analysis.

Edit - feeling a bit dim at the moment, and not really sure if any of the above was remotely useful or relevant. I think I'll step away from the PC for a bit!

It's possible to design such system so that markers can be added by users, but yes there will be a lot of data in this system.

 

[Leopardtail]

Perfect Esther - you added some things in there that I hadn't thought of. I'm getting a Wiki set up on the Forum that we can use to work on this stuff more easily. thanks!

Again Cort, this comes back to the 'effective design' discussed earlier.
One obvious thing here: my symptoms vary MASSIVELY with fatigue levels as I suspect would my blood results, these patterns are I suspect crucial to understanding the disease.