CFSME ATLAS site

The objectives are worthy.

My first concern is that we do not who is writing this (I don't think). The more I am involved with advocacy the more I think we should all say who we are.

The text i have looked at does not stand up to the standards aimed at. The pieces on What is ME/CFS?, PEM and Neuroinflammation all repeat very shaky claims.

I think it is bad policy to get ai involved in something like this. This is the perfect landscape for garbage in garbage out.

Sadly, I think the site is doing exactly what it sets out to avoid.
 
My first concern is that we do not who is writing this (I don't think).
There's an email at the bottom of the page indicating it could be Eivind Fasting.

When I clicked on the email, gmail automatically opened a draft to send to them, and the picture it showed by the recipient is one of the LinkedIn pictures that comes up on Google search with that name and surname. It's someone living in Oslo. Fb says we have an ME/LC group in common.

Full name based on my search is Eivind Lukas Moen Fasting.

If someone hasn't stolen their identity...
 
The website looks impressive but it seems like it's mostly AI summaries and AI evidence ratings of ME/CFS studies.

That can give impressive looking overviews and graphs but if nobody is doing the hard work of checking and evaluating results, then its quite the opposite of useful.
 
Last edited:
To give an example, it allows you to filter on symptoms such as post-exertional malaise.

It would be quite useful to quickly get an overview of studies that focus on PEM as its main topic. Unfortunately, this website probably uses mentions of 'post-exertional malaise' in the abstract.

This brain scan study "Distinct functional connectivity patterns in myalgic encephalomyelitis and long COVID patients during cognitive fatigue: a 7 Tesla task-fMRI study." isn't about post-exertional malaise at all, it just mentions it in listening ME/CFS symptoms. But it's included in the studies if you check post-exertional malaise.
 
Thank you for taking the time to look at the site, and for the criticism. I genuinely appreciate it.

A brief note on where this came from. I have been ill since 2015. Before that, I had never heard of ME/CFS. Once I became sick, I ran into the same problem many patients do: the information landscape was fragmented, the research was difficult to navigate, the language was often technical, and it was surprisingly hard to find clear overviews that were both serious and readable when one is cognitively limited.

Over the years I started collecting papers, links, notes, classifications, and summaries for my own use, simply because I needed a way to keep track of the field. CFSMEATLAS grew out of that private effort. The motivation was not to present myself as an authority, but to try to build something I had needed myself: a more structured, navigable, and transparent way into a very difficult research landscape.

The project is independent. I pay for it myself and I do the work myself. It is not backed by an institution, not funded by a company, and not intended as a substitute for specialist judgment, primary literature, or careful peer review. It is a new and evolving public knowledge project.

What I am trying to build is, in practical terms, an ME/CFS evidence atlas: a large structured research corpus, public study pages, topic pages, guide pages, and an evidence layer that tries to distinguish between what is established, what is emerging, and what remains uncertain. At the moment the site includes thousands of public structured study records and is built on a much larger ingestion layer behind the scenes. The ambition is not simply to collect volume, but to make the field more legible without pretending it is more settled than it is.

That said, I think some of the criticism here is fair and useful. In particular, questions about trust, review depth, classification quality, PEM tagging, and the exact role of AI in summarisation and structuring are legitimate. I do not take those objections lightly. If anything, they point to the areas that need to become stricter, more transparent, and more editorially legible. My intention is not to let machine-generated structure masquerade as scientific certainty. The intention is to build a system that can become more robust over time, with clearer distinctions between machine-assisted processing, human review, and authored synthesis.

The broader vision is simple. ME/CFS patients should not have to reinvent their own knowledge infrastructure from scratch while severely ill. If this project becomes useful, I hope it can gradually become a calm, public, and trustworthy place where patients, families, clinicians, and researchers can orient themselves more easily in the field. But it is still early, and it will only improve if it is tested, challenged, and corrected.

So I am genuinely grateful for serious feedback, especially when it identifies concrete weaknesses. That is far more valuable to me than praise. The project is meant in good faith, and I would much rather make it stricter and better than defend weak points out of pride.
 
Seems that the person I described/"identified" is an architect, so that would explain the shaky claims you pointed out.
Yes, that is me.

I do not think my original profession is a meaningful argument here. The relevant question is whether specific claims, classifications, or summaries on the site are weak or overstated.

If they are, that criticism is fair and useful. But reducing the project to “he is an architect” just sidesteps the actual substance.
 
Trust-hardening pass — April 8, 2026

In response to criticism from the Science for ME forum — specifically that AI-generated structure was being mistaken for human scientific judgment, and that a brain scan study had been incorrectly tagged under PEM — we made the following changes:

Root cause fix The AI classifier was tagging studies as "PEM required" if PEM merely appeared in the abstract. We rewrote the classification prompt to be explicit: a YES classification requires PEM to have been a diagnostic entry criterion for participant inclusion, not just mentioned or measured. Criteria like Canadian Consensus, ICC, and IOM/SEID qualify; Fukuda and Oxford do not.

Label precision The filter label "PEM Status" was renamed to "PEM in Case Definition" with clearer option labels. The PEM badge on every study card now shows full-text ("PEM required" / "PEM not required" / "PEM unclear") with a tooltip explaining the distinction on hover.

Correction path Every study detail page now has a "Report an error" link that opens a pre-filled email to contact@cfsmeatlas.com with the study URL and title. Previously there was no way for a reader to flag a classification error.

Machine-draft transparency A visible notice was added to the evidence list page stating that most entries are machine-drafted and not yet human-reviewed, with a link to the methodology page. Previously this was only disclosed on the About page, which most users never reach.

Anchor link The "How the Atlas Works" section on the About page now has a proper anchor (#how-it-works) so study pages can link directly to the methodology explanation.
 
Hi @Robius welcome to the forum!

I think the general idea behind your project is really good. And the UI (while probably being LLM aided [i dont see that as a problem if the code is well audited]) is quite professional. The goals are definitely noble and you’ve identified a genuine gap.

I have major concerns over the scope of the project though. It seems like an impossible undertaking for one person. And relying too much on LLMs takes away from the rigour that would make such a project useful.

Also I want to say it’s great that you’re engaging here. But the messages you send sound very “LLMy”. If you’re using LLMs to like structure and format or translate your thoughts I have zero issue with that but some will likely be a bit taken aback.
 
I do not think my original profession is a meaningful argument here. The relevant question is whether specific claims, classifications, or summaries on the site are weak or overstated.

I agree that it is not a reason to reject your material. And you have made an impressive response to the S4ME fusillade!

I learnt about ME/CFSS for the first time in around 2015 as well but had the advantage of having been an academic physician working on chronic disabling disease for a lifetime before that. So being an architect you have had more ground to cover and you may not be familiar with just how bad much biomedical research and teaching is.

I was asked to co-chair an Invest in ME international workshop around 2015 and asked assembled scientists how much we were reasonably confident about in terms of biology of ME/CFS. They very honestly chorused "nothing". Individual researchers have to sell ideas to get funding and will come across as confident in findings but catch them for a moment being honest and they agree that we know almost nothing. We now have some gene linkages pretty much for certain but that is about it.

It might seem that to face up to this pervasive ignorance could be too painful to cope with but the S4ME community has shown that this need not be so. Progress is being made, even if most of the time it is exlcuding some other popular possibility that turns out to be a blind alley.

"We" don't think that ME/CFS involves neuroinflammation, in the sense that there is no reliable "we" view here. I don't think there is any inflammation and various members here have a range of views. And inflammation doesn't connect to low cerebral perfusion, which may or may not be relevant to ME/CFS.

All this information has to be interrogated the way a professional biomedical scientists would. That does not mean that architects cannot do that - we have people here of all sorts of backgrounds who interrogate data as well as any professional scientist. But nothing can be taken on trust.

You should join in the forum. It reflects the messy reality of uncertainty that is central to most biomedical problems. One bit of a study may be fine. Another bit may be a non starter. Some of the conclusions may be convincing. Others may not.
 
Hi @Robius welcome to the forum!

I think the general idea behind your project is really good. And the UI (while probably being LLM aided [i dont see that as a problem if the code is well audited]) is quite professional. The goals are definitely noble and you’ve identified a genuine gap.

I have major concerns over the scope of the project though. It seems like an impossible undertaking for one person. And relying too much on LLMs takes away from the rigour that would make such a project useful.

Also I want to say it’s great that you’re engaging here. But the messages you send sound very “LLMy”. If you’re using LLMs to like structure and format or translate your thoughts I have zero issue with that but some will likely be a bit taken aback.
Hi Yann04, sorry about that, yes I use Ai for writing sometimes, especially when it's this important to me or very technical, as english is not my native language. I follow Ai quite closely and I'm quite sure this work is within its capacity, though not perfect from day one. This is still very early in the project, and I'm learning and tweaking as we go. Your concern about the scope is natural, but in my world, this is what Ai is made for. An enormous databank of information which needs organisation and simplification. I understand that this is not for everyone though and can be quite concerning. Please come back to the site in a month or two and see if it has improved? Still have 5000 more papers to go through and had to cut down on the speed because of finances. (Written without ai :) )
 
Yes, that is me.

I do not think my original profession is a meaningful argument here. The relevant question is whether specific claims, classifications, or summaries on the site are weak or overstated.

If they are, that criticism is fair and useful. But reducing the project to “he is an architect” just sidesteps the actual substance.
Sorry, that comment was not intended to reduce the project to "he is an architect". It was merely a guess what could have contributed to publishing weak and problematic claims about ME/CFS without realising the issues around them. I don't expect anyone to know everything and be able to critically evaluate the whole body of ME/CFS research. If anything, I much more prefer when people acknowledge the gaps in their knowledge and ask questions.

I'm sure your transferable skills and interest in ME/CFS can get you somewhere, health permitting, and I appreciate your efforts.
 
Hi Yann04, sorry about that, yes I use Ai for writing sometimes, especially when it's this important to me or very technical, as english is not my native language. I follow Ai quite closely and I'm quite sure this work is within its capacity, though not perfect from day one. This is still very early in the project, and I'm learning and tweaking as we go. Your concern about the scope is natural, but in my world, this is what Ai is made for. An enormous databank of information which needs organisation and simplification. I understand that this is not for everyone though and can be quite concerning. Please come back to the site in a month or two and see if it has improved? Still have 5000 more papers to go through and had to cut down on the speed because of finances. (Written without ai :) )
Hahahah no worries. I have no issue with it. Just feel more comfortable understanding why using it and now that I do it helps. :)
 
I agree that it is not a reason to reject your material. And you have made an impressive response to the S4ME fusillade!

I learnt about ME/CFSS for the first time in around 2015 as well but had the advantage of having been an academic physician working on chronic disabling disease for a lifetime before that. So being an architect you have had more ground to cover and you may not be familiar with just how bad much biomedical research and teaching is.

I was asked to co-chair an Invest in ME international workshop around 2015 and asked assembled scientists how much we were reasonably confident about in terms of biology of ME/CFS. They very honestly chorused "nothing". Individual researchers have to sell ideas to get funding and will come across as confident in findings but catch them for a moment being honest and they agree that we know almost nothing. We now have some gene linkages pretty much for certain but that is about it.

It might seem that to face up to this pervasive ignorance could be too painful to cope with but the S4ME community has shown that this need not be so. Progress is being made, even if most of the time it is exlcuding some other popular possibility that turns out to be a blind alley.

"We" don't think that ME/CFS involves neuroinflammation, in the sense that there is no reliable "we" view here. I don't think there is any inflammation and various members here have a range of views. And inflammation doesn't connect to low cerebral perfusion, which may or may not be relevant to ME/CFS.

All this information has to be interrogated the way a professional biomedical scientists would. That does not mean that architects cannot do that - we have people here of all sorts of backgrounds who interrogate data as well as any professional scientist. But nothing can be taken on trust.

You should join in the forum. It reflects the messy reality of uncertainty that is central to most biomedical problems. One bit of a study may be fine. Another bit may be a non starter. Some of the conclusions may be convincing. Others may not.
Thanks Jonathan! As a Norwegian, I think we as a country might be behind on our official view on ME and that's partly why my science interest has grown. If the site does more harm then good I will not hesitate on pulling it down, but I want to work on it a few months and see where it ends up.
 
Welcome @Robius, I am not up to looking at your site today, but will come back to this later. I think this is a very valuable exercise.

A bit of a tangent, but I am struggling to read the name as intended. I struggle not to see it as CFS-meat-las. The misreading as the word ‘meat’ tends to override for me other interpretations.

Increasingly ME/CFS is being used. It may be others do not have this problem, but personally I would find MECFSAtlas a clearer format.
 
Back
Top Bottom