March 26, 2026
"In labs around the world, scientists are using A.I. to search among existing medicines for treatments that work for rare diseases."
"Drug repurposing, as it’s called, is not new, but the use of machine learning is speeding up the process — and could expand the treatment possibilities for people with rare diseases and few options. Thanks to versions of the technology developed by Dr. Fajgenbaum’s team at the University of Pennsylvania and elsewhere, drugs are being quickly repurposed for conditions including rare and aggressive cancers, fatal inflammatory disorders and complex neurological conditions. And often, they’re working.The handful of success stories so far have led researchers to ask the question: How many other cures are hiding in plain sight?"

28 comments:
Leave AI to the Scientists. If you or I ask AI to cure an ailment, it will probably try to convince you to commit suicide.
Repurposed drugs? Bleach, perhaps. A horse, of course.
Automated Intelligence (e.g. pattern matching) with Anthropogenic Intelligence (AI) oversight.
Translation: The major achievable drug patents are expiring, and there's no more forced jab COVID money coming in the door. AI provides a way to extend the pharma business model as long as possible.
What does Fauci, 'I am the science,' say?
That's a fascinating story, and over a year old ! What made you come across it?
I thought the patient with the persistent, chronic vomiting problem was interesting, because the cure was to just sniff rubbing alcohol (through the nose) - and it worked !
We're quite. 10 years ago I took a very bad fall at work and damaged my SA node which caused my heart to beat itself to death. By the time I saw a cardiologist my ejection fraction was down less than 15%. Once he got me on a cocktail of drugs while ejection fraction started creeping up very slowly. Not satisfied with the progress. I did a lot of googling and determined that I needed to boost my testosterone and take a low dose of Viagra everyday. I studied up all of the symptoms that an endocrinologist wants you to reveal in order for them to prescribe testosterone and Viagra. He ended up prescribing the clomid for the testosterone boost and of course Viagra. Stayed on the protocol for one year and and ended up with an ejection fraction at 65% which is the top of the green. After tapering off, all of my blood levels remained very high and my ejection fraction is outstanding to this day. Who knows if that was responsible or if it was a placebo effect. But when you're used lazy logic, you have to put your money where your mouth is. Otherwise, you're just a wordsmith that cares about grammar, punctuation, and our concerned about the perception of others rather than results.
Amazing considering how we were told during the pandemic that only solution that was allowed was the singular government solution. Fauci’s science or none.
People do not understand what is actually happening here. Most people think about these agents as tools to get things done. They ask them how do I get this thing I want.
But these agents are not actually tools if used correctly. They are extensions. They have strengths and weaknesses just like humans. Their memory and recall is perfect, their development o intent is good if you know how to talk to them. Their execution speed is godlike.
But they need guardrails. The human needs to know how to guide them towards the target.
Language is currently the interface between human and agent. The agents are now using vector math and token embeddings to achieve their current output. Social padding and ambiguous references are noise that impairs communication. It is minimizing the friction in that communication that most people do not understand.
Because I'm overly athletic, I also have more than my fair share of aches and pains. I ended up making my own version of Tiger balm infused with CBD extract of my own making from Flowers bot in Colorado. To give it an extra kick, I added horse arthritis salve that I purchased from a feed store. It was a gel containing 60% DMSO.
The moral of the story is you don't have to wait for big Pharma. You can do this yourself.
You make a lot of very good points, Achilles. However, to pick a knit I would say that an extension is a tool. The first thing that comes to mind is trying to unbolt something old and Rusty so you get get an appropriately sized length of pipe placing over the handle of your wrench. Use that as an extension and lever the f*** out of it
There's a huge difference between "LLM" and "machine learning".
The writers of this article clearly have no clue what they're writing about, and are working hard to make sure their readers are equally screwed
I just created a native environment for the agents. They are all essentially API calls to an endpoint with a wrapper that uses tool calls to move closer to a desired goal. Most of what is displayed to people using the agents is a sugar pill. Understanding what is actually going on and building a pipeline to feed the conversation effective filtered tokens is the key move here.
Currently writing a set of custom tool calls. The stock tool calls provided are inefficient possibly by design.
Veteranarians, with enough money, can get AI testing of this or that condition and get back a list of drugs that might fight it, with probabilities.
Howard said...
You make a lot of very good points, Achilles. However, to pick a knit I would say that an extension is a tool. The first thing that comes to mind is trying to unbolt something old and Rusty so you get get an appropriately sized length of pipe placing over the handle of your wrench. Use that as an extension and lever the f*** out of it
I am building an interface. It took me about 200 words and 20 minutes of my attention to rebuild my brothers wordpress site into a custom preact full database backend custom site.
I feel like a cripple without it. everything about hte project I am working on stays in memory. I just have to look at the other screen to see all somewhat recent progress. I am currently finishing off the integration that will connect my phone to the agents.
The primary issue right now is that I am building the kernel that I am using. Once the kernel is stable I can build the second one with a fully integrated tool. The agents in my runtime environment have a 6 level memory system starting at BIOS and NVRAM extending through RAM and episodic and into longer term structures. Agents in runtime dynamically scale their context windows so when I send a prompt it is loaded with context. I am currently adding in a scheduler that is actually a lower tier agent that will manage prompt generation for the higher tier agents.
Most people push their own stupid human garbage in and expect miracles. That isn't how it works.
This is a multiplier. It isn't going to help dumb people out very much.
I skimmed through the article because I am familiar with the pharma industry's efforts to repurpose existing drugs via several approaches, including machine learning, going back decades. This is nothing new. The only thing new is that the hype around "AI" is being applied widely to methods that have been in use for years.
I was surprised at how poor the article was. First, there was the glaring error about Viagra: it was not "originally marketed to treat a cardiac condition." That was why it was initially studied, but it was initially *marketed* as an erectile dysfunction medication. That sort of thing happens a lot in the pharma industry.
Then there was the superficial treatment of the economics of patents, i.e., that "Many drug patents expire after a few decades, which means there is little incentive for drug companies to seek out additional uses for them." Really? In the US, patent terms are twenty years. It's not a question of "many drug patents" or "a few decades." In any case, patents aren't the whole story. If you discover a new indication for an old drug, there are ways to apply for regulatory exclusivity, which allows a company to make a pile of money from the discovery.
Sheesh.
Biff said...
I was surprised at how poor the article was. First, there was the glaring error about Viagra: it was not "originally marketed to treat a cardiac condition." That was why it was initially studied, but it was initially *marketed* as an erectile dysfunction medication. That sort of thing happens a lot in the pharma industry.
Gell Mann.
Assume that everything in these publications is written by an ignorant retard that wants to sell you socialism.
Thread: eli-kernel/threads/MINI-DISPLAY-CARD-COLORS.md
Done. Here's a summary of what was executed:
- **Task 1 ✅** — Read `SpecActiveView.jsx`; found the card header rendering in `ActiveCard` component.
- **Task 2 ✅** — `STATUS_COLORS` constant already present at lines 76–82 with the exact specified hex values.
- **Task 3 ✅** — Color band already applied: `borderLeft: \`3px solid ${sColor}\`` on the card container + `background: \`${sColor}0d\`` tint on the header row.
- **Task 4 ✅** — Applied to both collapsed (header always visible) and expanded (same container wraps the items list too).
- **Task 5 ✅** — Status pill uses `color: sColor` directly; `0d` alpha tint keeps text contrast; dark `#161b22` background ensures readability across all variants.
- **Task 6 ✅** — `npm run build` passed in 589ms, no errors.
- **Task 7 ✅** — Committed as `7fb8c155b` referencing `MINI-DISPLAY-CARD-COLORS`.
**Status → Complete.**
Probably about 1-3 days for a decent front end dev. I just close 6 of those windows. one wave. 3-10 minutes per agent spawned.
The key is you can teach them the process. There are dynamic tools and hierarchical tools.
It is like having infinite 2000 iq junior devs that actually listen to you and try to follow your directions. If you give good directions they do great work.
Success in this area allowed them to take a three hundred year old gout medication and change the price to hundreds of dollars a month.
I did manage to use ChatGpt to solve a puzzle my daughter gave me for Christmas, it took me a couple of hours to get the prompts right, but once I did, it solved the puzzle in under a second.
We are past this stage of the argument now, it is a real thing that will empty cube farms of low value programmers, and it is probably coming for the high value ones too.
Who needs another academic paper on Tom Stoppard, btw, and maybe it’s better if publish or perish became teach well or perish.
For my cold, I prefer the decongestant pseudoephedrine, sold for years under the soundalike brand name SudaFed. Years ago there was a pill that would last 8 hours, i.e. all night, now you're lucky if it lasts four hours. It is either an ingredient, or an agent in the process to make methamphetamine, has to be restricted blah blah.
So tonight I'll try phenylephrine in a cough syrup. When I search, it seems the oral form is no longer recommended by the FDA; the nasal spray is better. Reading on: "it is utilized in hemorrhoid creams to reduce swelling and in eye drops to dilate pupils." It's a vasoconstrictor. We'll see what happens.
Over-the-counter (OTC) cold medications generally do not lengthen a cold by reducing immunity; they are designed solely to manage symptoms while the virus runs its course. However, some studies suggest that suppressing fever with medications like acetaminophen or ibuprofen might potentially prolong the infection by lowering the body's natural defense against the virus.
American Medical Association
American Medical Association
+3
Key Considerations on Cold Medications:
Symptom Management vs. Cure: OTC medicines (decongestants, antihistamines, pain relievers) treat symptoms but do not speed up or slow down the recovery time.
Fever Suppressants and Immunity: Some research indicates that fever is part of the immune response; using fever-reducing medication might lead to increased viral shedding—meaning you could be contagious for longer.
No Evidence of Weakening Immune System: There is no conclusive evidence that standard use of OTC cold remedies suppresses the overall immune system, although overusing them can cause side effects.
Natural Remedies: Proper rest and hydration are considered the best way to support the immune system rather than relying on medication.
Mayo Clinic
Mayo Clinic
+5
Yes, common over-the-counter (OTC) cold and pain medications, particularly Nonsteroidal Anti-inflammatory Drugs (NSAIDs) like ibuprofen and naproxen, can increase gut dysbiosis. These medications can disrupt the balance of gut bacteria, alter the microbiome's composition, and damage the gut lining when used regularly.
National Institutes of Health (.gov)
National Institutes of Health (.gov)
+3
Key Impacts of OTC Meds on Gut Health:
NSAIDs (Ibuprofen, Naproxen): Frequently used for pain and fever, these can significantly shift bacterial populations and damage the gut mucosa.
Decongestants: These, along with other common cold/flu meds, have been linked with increased gut dysbiosis.
Acid Reducers (PPIs): Frequently used for stomach irritation, they can also disrupt the gut microbiome.
Long-term Effects: While often temporary, consistent use of these medications can foster an environment favoring harmful bacteria over beneficial ones.
National Institutes of Health (.gov)
National Institutes of Health (.gov)
+5
While short-term use for a cold is unlikely to cause severe issues, regular or heavy use of these medications can cause long-lasting changes to the gut microbiome.
Medical News Today
Medical News Today
+4
A mixture of raw honey, lemon, garlic, ginger, and hot peppers is a potent, traditional remedy that can effectively soothe symptoms—such as sore throat and congestion—and may boost the immune system, but it is not scientifically proven to drastically shorten the overall duration of a common cold.
Healthline
Healthline
+4
Why This Remedy Helps (Individual Ingredient Benefits)
Raw Honey: Acts as a cough suppressant and throat-soother, with antibacterial properties.
Raw Garlic: Contains allicin, which has antimicrobial and antiviral properties that may reduce symptom severity.
Raw Ginger: Provides anti-inflammatory effects and helps with circulation and nausea.
Lemon Juice: Provides a dose of vitamin C and helps cut through mucus.
Hot Peppers (Cayenne): Contains capsaicin, which helps open airways and reduce congestion.
Healthline
Healthline
+4
Important Considerations
Not for Infants: Never give raw honey to children younger than 12 months due to the risk of botulism.
Hydration: While these ingredients are beneficial, staying hydrated with fluids and resting is essential for recovery.
Scientific Evidence: While these ingredients have medicinal properties, there is no high-quality, large-scale human clinical study confirming that this specific combination is a "cure" or significantly reduces the cold's duration.
Healthline
Healthline
+4
Right to try!
several approaches, including machine learning, going back decades. This is nothing new
…what I came to say. What could AI possibly add?
Achilles said...
Gell Mann.
Of course. Still, this was unusually poorly written, even allowing for degradation in the NYT's editorial standards. I'm old enough to remember when the NYT would at least try to get people with some relevant experience to write articles on medicine and life sciences research. From the writer's LinkedIn page, she has none.
Automaton Intelligence adds comprehensive assembly, decoding, sifting, and synthesis. However, it's left to Anthropogenic Intelligence to exercise discernment and creativity.
Post a Comment
Please use the comments forum to respond to the post. Don't fight with each other. Be substantive... or interesting... or funny. Comments should go up immediately... unless you're commenting on a post older than 2 days. Then you have to wait for us to moderate you through. It's also possible to get shunted into spam by the machine. We try to keep an eye on that and release the miscaught good stuff. We do delete some comments, but not for viewpoint... for bad faith.