AI

[ruben]

Keir Starmer has just said he will make AI work for everyone. Will this include ME/CFS sufferers I wonder and how quickly

 

[Wayne]

If pwME-CFS aren't already using AI in some way to help themselves, then I think they're missing out on a powerful tool. I have little hope any kind of "agency" will help us in some miraculous way using AI. But I do believe AI resources allows us to help ourselves in a myriad number of ways, which cumulatively, can make a meaningful difference for many of us.

 

[hapl808]

If pwME-CFS aren't already using AI in some way to help themselves, then I think they're missing out on a powerful tool.

It's probably more helpful than any doctor I've seen in over 20 years, other than the lack of a prescription pad.

 

[linusbert]

yea if ai could give out prescriptions i would not need to go to any doctor anymore.
also most psychatriests would go unemployed...

 

[Carl]

Knowing what I know, I don't believe that AI will be capable of finding or working out the cause of ME/Fibro because it can only draw on current knowledge and that is where it will fail. The required knowledge is so disconnected that it will require creative thinking which I don't think AI will be capable of doing. Connecting information unknown to medicine about the workings of the human nervous system and bacteria which is something that I can guarantee AI will not be capable of doing any time soon. By all means come back to this thread and tell me I was wrong but I don't believe that is likely to happen TBH. Current knowledge about one bacteria still hasn't made the required connection and the necessary knowledge is not known.

It will take a human with the correct thinking to work it out. How long it takes will depend upon events and the way things are going that could be a very long time.

 

[bad1080]

don't forget current AI isn't intelligent at all, it can only regurgitate what's already out there. plus it will do errors while doing it and it will confidently tell you the wrong thing and double down if pressed on the issue. you'd have to double check everything it spits out, so i don't see the benefit in using it if it ends up being more work.

https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)

 

[ruben]

If pwME-CFS aren't already using AI in some way to help themselves, then I think they're missing out on a powerful tool. I have little hope any kind of "agency" will help us in some miraculous way using AI. But I do believe AI resources allows us to help ourselves in a myriad number of ways, which cumulatively, can make a meaningful difference for many of us.

There must be many out there who've had ME/CFS for many years, including me. Some might not even be especially tech savvy. Where the heck do they start from the point of view of AI helping their ME/CFS

 

[pamojja]

Where the heck do they start from the point of view of AI helping their ME/CFS

When I started my health Odyssey many years ago - first with a walking-disability from PAD, next a COPD and last constant PEMs - the biggest starting point from me was really learning everything pertinent to so many possibly helpful supplements, and about not less laboratory tests (with which I monitored pro- or regress). Most learning wasn't classical from textbooks, but mainly in dialog form through asking, and soon answering (easier questions) and thereby consolidating or revising knowledge on various health forums myself.

Many years later, with the above in remission already for years again, I've forgotten many details again. Now where health-forums don't seem so lively anymore, I found AI a very good, though less controversial dialog partner, for basic relearning again. And just with health-forums, also with AI one better questions everything and ask for authentic sources. One tries as much in finding inconsistencies or mistakes of the AI, as one would with one's own,

from the point of view of AI helping their ME/CFS

At least in my case, I very soon didn't expect help, even from MDs. But through their means learned to help myself, wherever there seemed such a probability.

More tech-savvy can let it explore tons of underexposed science from the far past, which much has already been forgotten. By the focus to find synthetic patentable molecules for almost a century now.

 

[hapl808]

Knowing what I know, I don't believe that AI will be capable of finding or working out the cause of ME/Fibro because it can only draw on current knowledge and that is where it will fail. The required knowledge is so disconnected that it will require creative thinking which I don't think AI will be capable of doing. Connecting information unknown to medicine about the workings of the human nervous system and bacteria which is something that I can guarantee AI will not be capable of doing any time soon.

Are you basing this on vague theory, or practice? Because I hear this exact viewpoint all the time (discounting emergent properties, MoE, better post-training optimization, etc), and it's not my experience at all.

How come AI can easily and confidently diagnose illnesses that almost all doctors struggle to identify? EDS, MECFS, etc. If it were going by human training, you would think it would jump to anxiety, overtraining, deconditioning, etc. But it doesn't.

don't forget current AI isn't intelligent at all, it can only regurgitate what's already out there. plus it will do errors while doing it and it will confidently tell you the wrong thing and double down if pressed on the issue. you'd have to double check everything it spits out, so i don't see the benefit in using it if it ends up being more work.

https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)

It tends to be more accurate than 99% of doctors and researchers. Ask it a question about glial activation or immune pathways and show me where it hallucinates. Can they hallucinate? Absolutely. Can humans get things wrong? Sure. But a link to Wikipedia is less useful than showing me an example of harmful hallucinations in the field of chronic illness. If you want me to give examples of human harmful hallucinations in chronic illness, I can probably link you to 100 Twitter posts by physicians and researchers.

Is current AI going to solve our problems? No. But is it more helpful than 99% of the physicians you'll encounter? Most likely so.

 

[hapl808]

There must be many out there who've had ME/CFS for many years, including me. Some might not even be especially tech savvy. Where the heck do they start from the point of view of AI helping their ME/CFS

You can go to chat.com or claude.ai and just use a free account. Start off with asking simple questions - the same types of things you might pose on a forum.

If I think my MECFS might be related to neuroinflammation, what are several medications or supplements could I try and list them in order of safety (most risky first).

When addressing neuroinflammation potentially linked to ME/CFS, it’s important to proceed cautiously and consult with a healthcare professional. Below is a list of medications and supplements that may target neuroinflammation, organized roughly from most risky to least risky. Risks depend on individual factors such as dosage, duration of use, and pre-existing conditions.

Higher Risk Medications/Supplements​

1. Immunomodulators (e.g., Low-Dose Naltrexone - LDN)
   • Mechanism: Modulates immune response and may reduce neuroinflammation.
   • Risks: May cause vivid dreams, insomnia, or worsening symptoms in some; limited data on long-term use.
   • Safety: Generally considered low-risk in small doses (e.g., 1-4.5 mg), but start low and go slow.
2. JAK Inhibitors (e.g., Tofacitinib, Baricitinib)
   • Mechanism: Blocks inflammatory signaling pathways.
   • Risks: Suppresses immune system; increased risk of infections, blood clots, and cardiovascular events.
   • Safety: Only used under strict medical supervision; often experimental for ME/CFS.
3. Glucocorticoids (e.g., Prednisone)
   • Mechanism: Broadly suppresses inflammation, including neuroinflammation.
   • Risks: Long-term use can lead to immune suppression, osteoporosis, adrenal suppression, and weight gain.
   • Safety: Generally avoided for chronic conditions like ME/CFS due to risks.
4. IVIG (Intravenous Immunoglobulin)
   • Mechanism: Modulates the immune system and reduces inflammation.
   • Risks: Rare risks of severe allergic reactions, kidney damage, or blood clots; high cost.
   • Safety: Requires specialist supervision and may not be accessible or practical.

---

Moderate Risk Supplements/Medications​

1. Antivirals (e.g., Valacyclovir, Famciclovir)
   • Mechanism: Targets chronic viral infections that may trigger neuroinflammation.
   • Risks: Can affect liver and kidney function; side effects like headaches and GI distress.
   • Safety: Prescription only; safe if monitored.
2. Anti-inflammatory Pharmaceuticals (e.g., Celecoxib, Minocycline)
   • Mechanism: COX-2 inhibitors or other anti-inflammatory effects targeting microglia.
   • Risks: Gastrointestinal, cardiovascular risks (e.g., Celecoxib); potential antibiotic resistance (e.g., Minocycline).
   • Safety: Use with medical advice.
3. CBD/THC
   • Mechanism: Modulates endocannabinoid system, reduces inflammation, and may calm overactive nerves.
   • Risks: Psychoactive effects (THC); possible interactions with medications.
   • Safety: Low risk with proper dosage; choose high-quality products.

---

Lower Risk Supplements​

1. Omega-3 Fatty Acids (Fish Oil)
   • Mechanism: Anti-inflammatory effects, supports brain health.
   • Risks: Possible mild GI upset; bleeding risk at high doses.
   • Safety: Widely considered safe.
2. Curcumin (Turmeric Extract)
   • Mechanism: Potent anti-inflammatory, reduces microglial activation.
   • Risks: May cause GI discomfort; avoid if gallbladder issues or on anticoagulants.
   • Safety: Safe at recommended doses.
3. Luteolin
   • Mechanism: Flavonoid with anti-inflammatory and neuroprotective properties.
   • Risks: Mild GI upset.
   • Safety: Generally safe; found in supplements targeting mast cell activation.
4. Resveratrol
   • Mechanism: Modulates inflammation and oxidative stress.
   • Risks: May interact with blood thinners; some report mild side effects.
   • Safety: Low risk.
5. Alpha-Lipoic Acid (ALA)
   • Mechanism: Antioxidant that reduces inflammation and supports mitochondrial function.
   • Risks: Hypoglycemia in diabetics.
   • Safety: Safe for most people.
6. Magnesium (Magnesium Glycinate or L-Threonate)
   • Mechanism: Reduces nerve hyperexcitability and calms the nervous system.
   • Risks: Overdose can cause diarrhea or kidney issues.
   • Safety: Safe at recommended doses.
7. Vitamin D
   • Mechanism: Immune-modulating and anti-inflammatory.
   • Risks: Rare toxicity from excessive doses.
   • Safety: Safe when monitored.
8. Quercetin
   • Mechanism: Anti-inflammatory and antioxidant.
   • Risks: Few; possible GI upset at high doses.
   • Safety: Very safe.

---

Notes​

• Start with lower-risk options and monitor for effects.
• Avoid self-medicating with higher-risk medications like JAK inhibitors or steroids without medical supervision.
• Some supplements like curcumin and resveratrol may enhance each other’s effects when combined.
• Consider functional medicine testing for personalized guidance.

Now, you can jump on this and say "HEY! Maybe LDN should be listed after JAK even though it's in a category." And maybe my prompt was unclear (I meant list the meds in order, not the categories).

But this is my criticism of the AI-critics. We'll get an answer that is 90% amazing, and they'll try to pick at the 10%.

In 25 years of MECFS, I've never had a single doctor give me (or understand) a list like this. I already know all these, but that's from 25 years of research and forums and 2 years of using AI daily.

 

[bad1080]

It tends to be more accurate than 99% of doctors and researchers

idk how you got to this number but from my (admittedly very limited) contact with AI i came across hallucinations first hand. and when i pointed out the discrepancies the AI first doubled down on its mistake before spewing more nonsense. it simply isn't intelligent, nor is it aware of what it is doing.

Ask it a question about glial activation or immune pathways and show me where it hallucinates

well that's what i am saying: to be able to point out an error would include dealing with an AI, then doing all the research myself so i'd be able to know if it was correct in the first place. and i am not doing that but to each their own. just know the risks when you place your faith in an "AI".

 

[hapl808]

well that's what i am saying: to be able to point out an error would include dealing with an AI, then doing all the research myself so i'd be able to know if it was correct in the first place. and i am not doing that but to each their own. just know the risks when you place your faith in an "AI".

I'm definitely aware of the risks, but that goes with all sources. You included a link to Wikipedia which is a great resource, but like AI is far from infallible.

My own experience in not very limited contact with physicians over the past 25 years is I've gotten a ton of very harmful advice and treatments. Encouraging me to push through fatigue and cognitive issues, more PT than I could handle, rough exams causing me permanent injuries, etc.

I have found AI much more trustworthy in not very limited contact. YMMV.

I find most people talking about hallucinations tried GPT3.5 once and of course a primitive model couldn't handle sophisticated problems. But they usually have very limited experience with Sonnet 3.6 or Experimental 1206 or GPT4o and are unaware of the current level of AI. This is like someone saying that cell phone screens are terrible because they're using a Motorola flip phone.

I agree - you should treat ALL information with healthy caution. Get second opinions when dealing with doctors, double check Wikipedia entries, research what AI suggests, look into what someone on a forum says. Never accept things at face value, because nothing is 100% reliable.

But discounting AI as 'all hallucinations' is a waste. I've often asked people lately to give current examples of harmful hallucinations and they all have a similar answer, "I'm not going to spend the time to do that." Which means it's not easy to find. At best, they'll point to a vague study or someone else's claim, as opposed to a current response with a good model.

 

[bad1080]

You included a link to Wikipedia which is a great resource, but like AI is far from infallible.

i linked to wikipedia so people not aware of the issue can get an idea what i am talking about. and i am not treating wikipedia as the be-all, end-all but as a first glance into a topic.

research what AI suggests

but that's what i am saying, it defeats the purpose. all of a sudden it is not easier, it's being more complicated and it ends up taking longer. it's like studying a topic by grading papers, which to be fair is an approach but certainly not my preferred one.

discounting AI as 'all hallucinations'

i never did that. the problem is you don't know when and where it will produce errors so you have to take all of it with tons of salt.

 

[pamojja]

I find most people talking about hallucinations tried GPT3.5 once and of course a primitive model couldn't handle sophisticated problems.

Yea me too. In the beginning, I couldn't believe how stupid, more stupid than me drying unsuccessfully teaching it to count (ie. 1,2,3..). Tried later again with a question about statics of a build construction. There I found definitive differences between different AI iterations. Since I knew from practical applications what different dimension could bear, it was very easy to spot those models prone to many hallucinations. So with expertise in one field of knowledge, the most errorful AIs are easy to spot and avoid.

just know the risks when you place your faith in an "AI".

That's why I said, instead, place your critical thought in an AI. And personal learning is actually initiated.

Aside from our intractable, complex questions about ME/CFS. Without being able to spot AI and somehow even beat it, Humanity anyway won't survive in the very near future. Manipulated in all aspects of life to the secret business models of countless billionaires, sucking humanity and its natural resources for cash.

Or should I say data? - the new currency. Only large enough computing power in the hands of a few is able to handle? To social engineer the perfect exploitable human. Including the dystopian consensus, to dispose 'unworthy' life.

Those who fall for opinions nowadays without critical questioning, will then, through more allpervaisive AI, even more certain.

 

[hapl808]

but that's what i am saying, it defeats the purpose. all of a sudden it is not easier, it's being more complicated and it ends up taking longer. it's like studying a topic by grading papers, which to be fair is an approach but certainly not my preferred one.

I guess I just fundamentally view things in a different way. I find AI much, much faster to start off research. For instance, rather than look up MECFS immune pathways at a high level, I can immediately jump into specifics to inform my research.

If I Google "immune pathways MECFS", this is the first result:

Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) involves debilitating fatigue, intolerance to exercise, and cognitive problems. Symptoms tend to get worse after even mild exertion. The cause remains unknown, hampering development of treatments. ME/CFS often develops after an infection—a condition known as post-infectious-ME/CFS (PI-ME/CFS).

A research team led by NIH’s Dr. Avindra Nath conducted an in-depth study of 17 people with PI-ME/CFS and 21 healthy volunteers. Their aim was to uncover key features that might be driving PI-ME/CFS. Results appeared in Nature Communications on February 21, 2024.

People with PI-ME/CFS had higher heart rates throughout the day and a smaller drop in their nighttime heart rate than healthy volunteers. This suggests that the autonomic nervous system, which controls unconscious bodily functions, is disrupted in PI-ME/CFS. Heart and lung function were also less able to respond to exercise in people with PI-ME/CFS.

In one test, participants were repeatedly given the choice of performing either an easy task for a low reward or a hard task for a higher reward. Those with PI-ME/CFS were less likely to choose the hard task than the healthy volunteers, suggesting less ability to exert effort. The healthy volunteers had reduced muscle function and brain motor cortex activity after repeated tasks. But people with PI-ME/CFS did not. Brain scans during one task found that those with PI-ME/CFS had lower activity in a brain region called the temporoparietal junction. These findings suggest that the fatigue of those with PI-ME/CFS might be caused by dysfunction in the way the brain decides how to exert effort.

When the team analyzed cerebrospinal fluid, they found that participants with PI-ME/CFS had reduced levels of chemicals called catechols. Catechols help regulate the nervous system. Catechol levels correlated with effort preference and motor function in people with PI-ME/CFS, but not in healthy volunteers. This suggests that altered catechol signaling in the brains of people with PI-ME/CFS may give rise to their altered effort.

The team examined immune function as well. They found differences in B cells, which make antibodies to help fight pathogens. People with PI-ME/CFS had more naïve B cells, which can be activated by any foreign substance. But they had fewer switched memory B cells, which respond to a specific pathogen that the body has encountered before. B cell dysfunction was more prominent in women. These findings suggests that the immune system continues to be activated in the absence of infection.

Taken together, the results suggest that PI-ME/CFS may be caused by immune system dysfunction that is triggered by infections. This may lead to chemical changes in the central nervous system that affect certain brain functions to cause ME/CFS symptoms.

“Rather than physical exhaustion or a lack of motivation,” says first author Dr. Brian Walitt, “fatigue may arise from a mismatch between what someone thinks they can achieve and what their bodies perform.”

“Men and women were quite divergent in their data,” Nath notes, “and that tells you that ME/CFS is not one-size-fits-all. Considering male and female immune differences in ME/CFS, the results may open up new avenues of research that could provide insight into other infection-associated chronic diseases.”

The findings also suggest ways in which PI-ME/CFS might be treated, either by targeting the immune system or specific circuits in the brain.

—by Brian Doctrow, Ph.D.

If I ask AI to explain in simple terms what specific immune pathways might be involved with MECFS, this is the result:

Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a complex and debilitating disease with an unknown cause. However, many researchers believe that a dysfunctional immune system plays a significant role. Here's a breakdown of the hypothesized immune pathways involved, explained in simple terms:

1. Chronic Inflammation:

• What it means: Think of inflammation as your body's alarm system that goes off when there's an injury or infection. In ME/CFS, this alarm system seems to be constantly "on," even when there's no obvious threat.
• How it's involved:
  • Cytokines gone wild: Cytokines are like messenger molecules that tell immune cells what to do. In ME/CFS, there can be imbalances in these cytokines, with some (pro-inflammatory cytokines) being overactive, contributing to ongoing inflammation.
  • Persistent activation: The immune system appears to be stuck in a state of heightened alert, even in the absence of a current infection. This could be due to a past infection that triggered an overreaction, or a failure of the system to properly "switch off" after an infection.

2. Natural Killer (NK) Cell Dysfunction:

• What it means: NK cells are like the body's special forces that quickly identify and destroy infected or cancerous cells.
• How it's involved:
  • Reduced activity: Studies have found that NK cells in many ME/CFS patients don't function as well as they should. They might be less able to kill target cells, which could make individuals more susceptible to infections or make it harder to clear them.
  • Impaired development: There might be issues with how NK cells develop and mature in ME/CFS, affecting their overall effectiveness.

3. Autoimmunity:

• What it means: Autoimmunity is when the immune system mistakenly attacks the body's own healthy tissues.
• How it's involved:
  • Autoantibodies: Some researchers have found autoantibodies in ME/CFS patients. These are antibodies that target the body's own cells and tissues, potentially causing damage and contributing to symptoms.
  • Misdirected attack: The immune system might be reacting to a perceived threat (like a past infection) and mistakenly targeting normal cells, leading to ongoing inflammation and tissue damage.

4. T Cell Abnormalities:

• What it means: T cells are another important part of the immune system. They are involved in recognizing specific threats and coordinating immune responses. There are different kinds of T cells, including "helper" T cells (which activate other immune cells) and "killer" T cells (which directly destroy infected cells).
• How it's involved:
  • Exhaustion: T cells in ME/CFS patients can show signs of exhaustion, meaning they are less responsive and less effective at fighting infections.
  • Skewed responses: The balance between different types of T cells might be disrupted, leading to an inappropriate or ineffective immune response.

5. Viral Reactivation:

• What it means: Some viruses, like the Epstein-Barr virus (EBV, which causes mono), can stay dormant in the body after an infection. If the immune system is weakened, these viruses can reactivate.
• How it's involved:
  • Trigger or consequence: Some researchers believe that viral reactivation might be a trigger for ME/CFS in some individuals. Alternatively, it could be a consequence of the immune dysfunction already present in the disease. Either way, viral reactivation could contribute to symptoms by causing ongoing inflammation and immune system activation.

Important Notes:

• Not all patients are the same: ME/CFS is a heterogeneous disease, meaning it affects people differently. Not everyone with ME/CFS will show the same immune abnormalities.
• Cause vs. effect: It's still unclear whether these immune problems are the primary cause of ME/CFS or a consequence of the disease process. It is likely a combination of both.
• Research is ongoing: Scientists are actively investigating these immune pathways to better understand their role in ME/CFS and to develop targeted treatments.

In summary, ME/CFS is likely associated with a dysregulated immune system, characterized by chronic inflammation, problems with NK cells and T cells, potential autoimmunity, and possible viral reactivation. These immune abnormalities could contribute to the wide range of symptoms experienced by individuals with ME/CFS.

I know which result I find easier. Now, if I Google and read the first 20-30 hits/papers, then I might get to a similar level of understanding (if my concentration lasts that long, which it likely will not).

 

[linusbert]

idk how you got to this number but from my (admittedly very limited) contact with AI i came across hallucinations first hand. and when i pointed out the discrepancies the AI first doubled down on its mistake before spewing more nonsense. it simply isn't intelligent, nor is it aware of what it is doing.

well you go to 10 doctors, 9 tell you outright you are psychotic... the other one isnt sure yet, but will jump bandwagon from the first 9 doctors because 9 other doctors cant be wrong, right?

 

[hapl808]

well you go to 10 doctors, 9 tell you outright you are psychotic... the other one isnt sure yet, but will jump bandwagon from the first 9 doctors because 9 other doctors cant be wrong, right?

In all fairness, those doctors simply aren't intelligent, nor are they aware of what they are doing to us.

 

[bad1080]

idk how terrible doctors are an argument for unreliable AI, seems like two separate problems to me.
how can an AI be more reliable than the research it bases its answers on? https://venturebeat.com/business/wh...guments-for-abolishing-the-black-box-vb-live/
the AI "inherits" all the biases ingrained in its code and from the source material it was trained on. AI still regularly recommends CBT and GET for me-cfs, i fail to see how this is better than the terrible doctors nobody would recommend visiting.
i am going to add just one more thing to this: an AI "quoted" a "source" (with DOI number!) for an answer it created but when looked up the DOI was nowhere to be found. so at a glance it looks legit but once you look into it (and only if you actually decide to look into it, which is hard enough as is for those with limited energy) it all falls apart. it is actually just hallucinating stuff but doing so with confidence, including creating fake sources when necessary (or rather when convenient).

 

[pamojja]

All shortcomings aside, AI already now helps greatly finding specific answers.

Till now, I had to do an internet search to find an answer. I had to read many articles, only to find, most are commercial, or didn't answer my specific question. Hours of often fruitless efforts. And PEM accordingly.

Now with AI one gets an often very intelligent summary right away. And in most cases one has to verify 1 article in detail only. No comparison.

If an AI fabricates non-existing sources to answer my specific question, it lost me as its user. I go to the next better one.

 

[kushami]

I have found it useful in the same way as @pamojja. I ask it a question, like “Can taking steroids affect creatine kinase blood tests?” and then verify what it tells me.

Even with the fact checking, it’s still much quicker than a PubMed search for “steroids creatine kinase” because that brings up every article that happens to mention the two, even if they aren’t connected in the article.

I wouldn’t call anything I do proper research, but I do often need to look up interactions, or find out whether a certain medication has a vasodilating or vasconstricting effect, or whether thee’s ever been a study on a substance to see whether it affects cerebral blood flow.

I hate making accounts for things, even free things, so I have been using Perplexity, thanks to someone on this forum who said it was pretty good. I think it will show sources if asked.