Use of EEfRT in the NIH study: Deep phenotyping of PI-ME/CFS, 2024, Walitt et al

Maybe they weren't pacing properly because they were confused by the instructions.

 

and this is probably the important bit to note because probably all, but certainly the iller ME-CFS would be in rolling PEM when doing the task and the fatiguability can be increased in PEM as well as other symptoms too. The 'rolling part' meaning that you get 'immediate hurt' from exerting in PEM and you are 'adding to the bill' in an exponential-feeling way by finding work-arounds to try and do what you need to do. It is hopping on a broken leg. Except you hopped on it for the last 5 days too?

 

This script is in no way perfect, but if any also want to request to amend 'effort preference' here was my high-level message with asking. Feel free to use any parts of it for those who want to reach out as well

https://docs.google.com/document/d/17LFcZmrIK4KNS7vZkjgGnnji9HzOrUQ0roIH-Qr053E/edit

 

Has anyone else got in touch to help tackle the dredging through the table of papers that use the EEfRT to answer these questions? And are there other questions we have?

Would it be helpful if we divided them up somehow (either by questions we are looking for the answers to, or by a group of papers) to do this? Or just a second pair of eyes - I don't know what I can promise on 'how many' (and quality might drop off with amount) but if someone is working top-down and the other bottom-up it might compensate a bit?

Happy if you'd feel better giving explicit instructions on how we do it (whether first it is a quick scan to see if it is touched on to narrow down into batches or what) given I might not be as good as you are/miss things you might have picked up?

 

Interesting chart. I'm trying to wrap my brain around without strategy how the pure maths would operate here:

OK in conclusion there is no way of even guessing this out. SO any approx heuristic is enough. And at the end of the day ... even if people could win $8 (which I'm pretty sure they can't without defying probability on the 'pick 2') vs $2 which is the probable outcome if you were to pick 55% of the easy ones that 'counted'...



Here's what probably really matters:

I'll contextualise. I used to run focus groups.

Over a decade ago (when money was worth more) a youngster with not much money and a lot of energy would say in response to £10 EDIT: (as an incentive to recruit participants to said groups) that you couldn't buy a magazine with that. and they would be correct.

In fact the hourly rate in a job is now near that. And this was validated on undergraduate students. Who may not be taxed ? (depends on the country I guess?) and so I don't know what the other amount for doing the trial itself was but $8 even vs $2 = $6 difference


Although I doubt many would have clocked the exact reward magnitude for one is supposedly being used as a dangling carrot?

It would certainly be interesting to read some of the raw data on the papers from eg Ohmann where they interviewed participants to see what they thought of it and what it was about..

 

I'll take a look!

I'd be up for taking on a portion if it'd be helpful.

 

I'm tagging @EndME

 

You'd think if it is amphetamine as I'm assuming it is ('speed') and the 'd' doesn't make it something different that both of these effects are both relevant and expected and part of the 'normal response curve' if a population were given that drug? thrill-seeking or whatever behaviour and feeling particularly 'energised' for doing lots of button-pressing?

 

I'm just adding this as it is an update to the HV vs ME-CFS No hard selected by level of probability, this time with HVF excluded (I've done a table to the left to show figures if they were included). These are just crude numbers and % without the extra variables included. I haven't looked to see if there are other things that were 'invalid' that I needed to have excluded. If we wanted to send anything off anywhere then I'd need to do a version that was more checked, as this is more quickly done to give a sense and as I'm exhausted I'm not beyond having made an error (or two)




No hard/person ME/CFS 0.5 level: 5.53
No hard/person ME/CFS 0.88 level: 10.53
both med+high level: 16.07

No hard/person HV 0.5 level: 7.44
No hard/person HV 0.88 level: 10.56
both med+high level: 18

So excluding HVF the difference is still particular at the 50:50 level of probability and is now nearer to 2 hard choices per person difference at that level (it was around 1.5 when HVF was included).

Although there is now slightly more of a % hard chosen difference between HV (64.75%) and ME-CFS (62.95%) , and still a 2% difference at the low probabililty level - with HV picking hard 2% more there too, the difference really does seem to still sit in that 50:50 probability area.

Which of course makes sense given the way the game works that would be where variation would logically take place (particularly if you were 'handicapped' in the metaphorical sense, so had to focus your choices of hard due to capability limitations)

 

OK I spoke to soon of course. I forgot about the warm-up rounds. SO the following tables exclude the warm-ups/trials with a '- number' ANd also excludes HVF (with the info if they had been included on the right)



Which make for the following calcs by 'person' (15 ME-CFS participants, 16 HVs):

No hard/person ME/CFS 0.5 level: 4.67
No hard/person ME/CFS 0.88 level: 10.00
both med+highlevel: 14.67

No hard/person HV 0.5 level: 6.38
No hard/person HV 0.88 level: 10.13
both med+high level: 16.5

This actually brings out what looks like quite a big % difference at the low-probability level (hard chosen 12.55% by ME-CFS and 18.41% by HVs at this level), even though the numbers behind it are actually small (HV chose 44 hard, ME-CFS 29 = 15). It equates to around 1 more hard choice per person though?

And a 2.5% difference at the high probability level, but underneath that is only 12 hard choices.

I've done the calculations per person because of the difference in 'N' number of participants in each group. At the high probability level, the difference in number of hard selected per person is 10 (ME-CFS) vs 10.13

At the 50:50 probability level it is 1.71 hard choices per person difference: 4.67 were chosen as 'hard' for ME-CFS and 6.38 for HVs (out of around an average of 16 tasks at that level to choose either easy or hard from)

 

another spot that is puzzling me (always a chance it is a data error) is that in total HVs did 734 tasks vs ME-CFS 707. EDIT: scrap that (as each person did around 40 tasks then it starts dropping off if they did lots of hards) That is more than what is accounted for in the different number of participants in each group (16 vs 15).

Treadway et al (2009) noted I think that they cut off data at trial 50, although they ran their test for 20mins and so I think that meant all participants were still playing.

Does anyone know whether something similar was used by Walitt et al (2024)?

This makes an average for HVs of round 45.875 and ME-CFS 47.1 rounds ie one more round on average - although I can see from just scrolling the big table I did that obviously wasn't evenly spread.

I think it is mainly important for data analysis purposes to see if I need to cut out later trials if Walitt has used a cut-off like Treadway did?

 

Great find, @bobbler . Right off the bat we have what may be the dealbreaker - Treadway recommended individual calibration of what constitutes a hard task in this schizophrenia study:

If I'm repeating things others have already said, oops. My individual calibration for S4ME is lower than some others'!

 

And from the discussion (Reddy 2015, https://academic.oup.com/schizophreniabulletin/article/41/5/1045/1921437?login=false#83746476):

 

A bit of listening from the NIH would go a long way, certainly on the EEfRT

a) they got the EEfRT "effort preference" research wrong.
b) the term sucks and doesn't describe ME. Some of the brain and hand grip findings are interesting, in my view. But why use "effort preference", which the MEAction response piece suggests is a term from mental health? Preference is insulting, as is the paper's claim that patients were trying to avoid "discomfort". They have no idea.
c) They don't seem to have a good grasp of how ME affects people. E.g. the battery of tests over a week for a patient cohort with an average SF36 phys function score of 30.

Perhaps it's time to listen to patients, or even partner with them (DecodeME) and/or learn from Patient-led Reserach?

It would probably lead to better science and less misunderstanding.

 

EDIT: added latest table which hopefully is easier to look at colour-wise. I lost the old quote stuff, so I'm quoting myself from the old post in the first bit but removed things I've now changed as it was a work in progress with rows missing then etc (apologies for any repetition)

OK, a new version of this table I posted before the weekend, because there was an error for HVO (that actually being a column copied over column from HVN by accident) and it included HVF. But also I've tried to clean it up a bit (!), changed the colours

I've also ordered within cohorts (HV and ME-CFS) by the number of completed hard tasks. To see if it helps to see any 'overlap/similarity' between HVs and ME-CFS who are doing better. But also hopefully to put the 'middling/in-betweener' ME-CFS in the middle of that cohort, between those who are similar in choices to HVs and those who clearly have capability-issues with the hard task and number of clicks (which is the number next to the coloured square so you can see how far off some were from the 98 to complete).




if anyone is just glancing at it then

each coloured square is a hard task - blue for where the participant completed it and pink for when they didn’t.

the participants are in columns so you ‘read down’ to see the pattern of complete/non-complete of their chosen hard tasks and obviously the white/blank is where they chose easy. So hopefully gives a sense of how spaced out their choices of hard were.

all the HVs are in the left side (ordered by number of hard completed) all the ME-CFS are on the right (ordered the same way). So those who completed fewest hard tasks are on the far right.

next to these if you can see that close is number of clicks (to left of whether they completed or not) so you can zoom in and get a sense of whether they were far off eg 80 clicks or just missing eg 96, 97 when it is red.

one thing you should be able to see now as it is pretty stark is how ‘blue’ the HV dude is apart from one participant who ‘just missed’ completing quite a few hards at the start then got it together.

whereas the ME-CFS side has a lot of pink on the right hand side (almost down the middle)

when you look closely there are, I think, approx ‘groupings’ with around 5 who non-complete most and their clicks are generally not near-misses, then in the middle ones who fail to complete/have quite a few red but tend to be nearer-misses to 98 clicks and might have spaced things out more etc.

 

I've made a correction to one of my posts above. 47% of patients had a lower success rate for hard tasks than all healthy volunteers.

 

Great graph. Am sure I'm not the only one with this problem: I find charts/graphs really difficult to look at now. My brain seems to get dazzled by strong contrasts in colour. I find it difficult to look at stripes, dotted patterns (tiny dots can be OK)...geometric patterns are the worst, gah. The less the contrast, the better. We recently decorated the rooms I'm in and I hunted down paints with lower "reflectance values" and reduced the contrast between colours that were going to be next to each other. No bright white ceilings. Brain is grateful.

 

I looked at a few things you might expect to differ between groups if one group were having difficulty with the task. I hypothesised that a group having difficulty with doing a task would have:

• Longer choice time
• Lower recorded presses
• Higher completion time
• Lower button press rate

compared to the other group.

Here's how it looks (NB mistakes possible, so double check if using)



It seems to me that lower # recorded presses, higher completion time and lower button press rate for pwME in hard tasks could support the argument that they found hard tasks more difficult than healthy volunteers, if these differences are statistically significant differences. Anyone able to check if these are statistically significant differences?


The paper states

which I understand to be a check for fatigue - if pwME were fatiguing quicker during the task than healthy volunteers, then they would have found a difference here. But what I’m talking about above with lower button press rate is not pwME becoming less capable over the course of the task, but potentially being less capable to begin with.

 

I did think some mightnt be able to and nearly wrote the words magic eye etc in the ‘whether people find it useful’ etc but then conscious of rambling on

I’m assuming it wouldn’t and don’t want to minimise but if any changes like me ‘going pastel’ on colours or playing with lines would make it more accessible to you then let me know x happy to give it a go etc if it makes a difference to anyone x

totally understand and I’m hoping it doesn’t dazzle anyone with how I’ve posted it as they scroll though?