AI in medicine, explained for everyoneGeneral information only, not medical advice · Our standards

Drug discovery · Drug discovery

An AI matched old drugs to a rare disease, and doctors credit it with helping one man

How it was checked
AI-assisted source check. Not clinically reviewed. ()
Last updated
Next review
Page ID
TR-045

General information only, not medical advice. Talk to your doctor about your situation. Read the disclaimer.

Researchers at Every Cure, a nonprofit, use AI to score how likely each of about 4,000 existing drugs is to treat each of about 18,500 diseases. Its doctors say the AI's top picks helped a man with POEMS syndrome, a rare blood disorder, who was preparing for hospice; he had been in remission for 9 months when the authors wrote it. It's one unpublished case that hasn't been independently checked. Rating: Announced only.

How far away is this?

A commentary in The Lancet Haematology (February 2025), not a study. The patient story is one case marked as unpublished data, so it's Announced only. It was published in early 2025, and we haven't found a later published report of the case.

Where is it approved?

RegionRegulatorStatusNotes
United StatesFDANot applicableNot a medical product; doctors prescribed already-approved drugs off-label.
European UnionCE mark / EMANot applicableNot a medical product; doctors prescribed already-approved drugs off-label.
United KingdomMHRANot applicableNot a medical product; doctors prescribed already-approved drugs off-label.
AustraliaTGANot applicableNot a medical product; doctors prescribed already-approved drugs off-label.

The AI platform isn't a regulated medical device; the drugs it points to are already approved for other uses. "Register not searched yet" means we haven't searched that regulator's register; it doesn't mean the product is or isn't available. Approval in one country isn't approval in another.

What is it?

Plenty of drugs work for more than one disease, but finding new uses usually happens by chance. Every Cure's platform links facts about genes, proteins, drugs and diseases into one big map, then uses machine learning to rank every drug-disease pair. The team, led by Dr David Fajgenbaum, who found a repurposed drug that saved his own life, wrote about it in The Lancet Haematology.

What's the evidence?

  1. The platform scores roughly 74 million drug-disease pairs (about 4,000 drugs times 18,500 diseases). Only about 4,000 diseases have an approved treatment, and the authors estimate over 300 million people have a disease with none.[1]

    Evidence stage: Announced only · #c1

  2. For a man with POEMS syndrome, a rare blood disorder, who had run out of treatments and was preparing for hospice, the team recommended several high-scoring drugs approved for multiple myeloma. The authors say these saved his life and he'd been in remission for 9 months when they wrote the piece; this hasn't been independently checked.[1]

    Evidence stage: Announced only · #c2

  3. The authors label the patient result 'unpublished data': it's one case, without the full details being shared. The work was partly funded by ARPA-H, the US health research agency.[1]

    Evidence stage: Announced only · #c3

What does it mean for me?

If you or someone you love has a rare disease with no good options, already-approved drugs are worth asking about, and AI tools like this may help doctors find them faster. One case doesn't prove the approach works. Any off-label drug should be a decision with your specialist.

What could I ask my doctor?

  • Are there drugs approved for other conditions that might help with mine?
  • Would you consider an off-label treatment, and what are the risks?

These are general starting points, not a recommendation to choose any particular test or treatment.

If you're worried about your health right now, contact your doctor or local health advice line. In an emergency, contact your local emergency number.

Change history

  • : First published.

General information only, not medical advice. Talk to your doctor about your situation. AI-assisted source check. Not clinically reviewed.

References

  1. The Lancet Haematology (2025): Pioneering a new field of computational pharmacophenomics. doi.org (2025). Expert commentary. https://doi.org/10.1016/S2352-3026(24)00278-3
  2. Author manuscript on PubMed Central (PMC12090018). pmc.ncbi.nlm.nih.gov. Expert commentary. https://pmc.ncbi.nlm.nih.gov/articles/PMC12090018/

Publisher shown as the website hosting each source. Dates are given where the source label includes them.