Solving ME/CFS website

Andy

Andy

Senior Member (Voting rights)
"ME/CFS is life-ruining, poorly understood and has no cure. An estimated 17–24 million people had it before COVID-19, and long COVID has added millions more. About a quarter are housebound or bedbound. The most severely ill cannot sit up, cannot bear light, sound or conversation, and need help to eat.

Solving ME/CFS exists to help solve the disease sooner. Our goals:
  1. Understand the disease. Build a useful model of the underlying mechanisms from all the evidence, graded by how well it holds up.
  2. Close the gaps. Get the research done that would answer the open questions and test the model, starting with the experiments below.
  3. Find treatments. Propose and test treatments aimed at those mechanisms."
....

"How it might work​

The current best explanation is a loop. A severe infection primes the immune system in people whose inherited risk lies in the brain's control of breathing, heart rate and blood volume (steps 1–2). Blood volume is low and vessels dilate poorly, so muscle and brain get too little blood when the body is upright or working (3–5). The under-supplied tissue may then be injured, setting off an immune response and, a day or two later, the sickness response patients call a crash (5–7). That part is untested in ME/CFS. Arousal and poor sleep then disturb step 2 again, so the illness can outlast the infection (8). The model page grades every step, lists the observations it explains poorly, and names the result that would disprove each step."

...

"Solving ME/CFS is written by Michael Jelly, with AI agents assisting with literature review and analysis."


Link to website
 
Seems like AI took all these papers at face value and fact then constructed a website around it. Ground truth for ME/CFS is so far off this approach for hypothesis generation with AI fails. This is like feeding Claude a bunch of papers you believe to be true then asking it to make a hypothesis around it, but it never questions the original papers. More noise with a wrapper imo
 
You must log in or register to reply here.
Back
Top Bottom