5 ms·

Minikanren was the subject of a phenomenal talk by Matt Might for it's applications in medicine [0] https://www.janestreet.com/tech-talks/algorithm-for-precisi

Minikanren was the subject of a phenomenal talk by Matt Might for it's applications in medicine [0] https://www.janestreet.com/tech-talks/algorithm-for-precision-medicine/ https://www.janestreet.com/tech-talks/algorithm-for-precisio...

Bluestein 2y ago
... wherein: "It will demonstrate progress to date, including the now-routine use of relational programming in miniKanren to identify personalized treatments for patients with some of the rarest and most challenging diseases in the world." ... wherefrom I cannot help but be amazed at how personalization through data optimization apparently actually helps solve such things and can be an approach.-
smeej 2y ago
A couple years ago, I tried to wrap my poor little non-developer head around how to use this, because I have a handful of things going on medically that seem (to me) to be connected, but I can't manage to get any doctors to care enough to help me figure out, but I was hopelessly out of my depth. Normally I can at least figure out how to use something, even if I don't understand how it works under the hood, but with this, I couldn't even figure out the next step.
- worldsayshi 2y ago
  You might have better luck if you try today using Claude as support?
  - will_byrd 2y ago
    Alas, you'd still need the knowledge graphs. I hope the licensing issues for at least some of the KGs will be resolved soon. It's a tricky issue. Even some of the ontologies and controlled vocabularies in biomedicine can't be released publicly due to copyright restrictions (for example, full SemMedDB uses UMLS which uses SNOMED).
    - worldsayshi 2y ago
      Makes me wonder if LLMs can be used to generate useful KGs (possibly from the right source material).
- will_byrd 2y ago
  Sorry about that. mediKanren's source code is under MIT license, and is on GitHub. Alas, the knowledge graphs we use for mediKanren aren't produced by us, and often have very complex licenses (one KG might include knowledge from 80 or 100 databases, each with a different license). As a result, we can't just release the KGs that are needed to actually use mediKanren. Also, creating high-quality queries that take into account all the nuances and quirks of the KGs is tricky, and changes as the KGs and the Biolink standard evolve. As a result, mediKanren requires some expertise to use effectively, along with access to the KGs. An application that uses mediKanren as a back-end (along with other reasoners) is the NIH NCATS Biomedical Data Translator: https://ui.transltr.io/ https://ui.transltr.io/ Please keep in mind that both Translator and mediKanren are designed for researchers and for biomedical research, not for patient care or treatment recommendations.
  - smeej 2y ago
    It's surprisingly reassuring to realize it wasn't actually my fault that I couldn't figure it out. Thanks for explaining! I'll look into Translator and see if I can get anywhere that way.
- sn9 2y ago
  Have you read their document [0] on how to navigate the process? [0] https://bertrand.might.net/articles/algorithm-for-precision-medicine/ https://bertrand.might.net/articles/algorithm-for-precision-...
isoprophlex 2y ago
That was an incredible read, a real-life Dr. House episode. And very exciting technology, too. Thanks for sharing.
riffraff 2y ago
fantastic talk, thanks for sharing.
dang 2y ago
Discussed a bit on HN: The Algorithm for Precision Medicine, talk by Matthew Might [video] - https://news.ycombinator.com/item?id=37814662 https://news.ycombinator.com/item?id=37814662 - Oct 2023 (14 comments) The Algorithm for Precision Medicine - https://news.ycombinator.com/item?id=29483549 https://news.ycombinator.com/item?id=29483549 - Dec 2021 (1 comment)
- fhe 2y ago
  thank you for posting this... and this is a common enough pattern (looking up previous HN threads on the same or related article) that I wonder if it's worthwhile to automate this.

Bluestein 2y ago

... wherein: "It will demonstrate progress to date, including the now-routine use of relational programming in miniKanren to identify personalized treatments for patients with some of the rarest and most challenging diseases in the world." ... wherefrom I cannot help but be amazed at how personalization through data optimization apparently actually helps solve such things and can be an approach.-

smeej 2y ago

A couple years ago, I tried to wrap my poor little non-developer head around how to use this, because I have a handful of things going on medically that seem (to me) to be connected, but I can't manage to get any doctors to care enough to help me figure out, but I was hopelessly out of my depth. Normally I can at least figure out how to use something, even if I don't understand how it works under the hood, but with this, I couldn't even figure out the next step.

worldsayshi 2y ago
You might have better luck if you try today using Claude as support?
- will_byrd 2y ago
  Alas, you'd still need the knowledge graphs. I hope the licensing issues for at least some of the KGs will be resolved soon. It's a tricky issue. Even some of the ontologies and controlled vocabularies in biomedicine can't be released publicly due to copyright restrictions (for example, full SemMedDB uses UMLS which uses SNOMED).
  - worldsayshi 2y ago
    Makes me wonder if LLMs can be used to generate useful KGs (possibly from the right source material).
will_byrd 2y ago
Sorry about that. mediKanren's source code is under MIT license, and is on GitHub. Alas, the knowledge graphs we use for mediKanren aren't produced by us, and often have very complex licenses (one KG might include knowledge from 80 or 100 databases, each with a different license). As a result, we can't just release the KGs that are needed to actually use mediKanren. Also, creating high-quality queries that take into account all the nuances and quirks of the KGs is tricky, and changes as the KGs and the Biolink standard evolve. As a result, mediKanren requires some expertise to use effectively, along with access to the KGs. An application that uses mediKanren as a back-end (along with other reasoners) is the NIH NCATS Biomedical Data Translator: https://ui.transltr.io/ https://ui.transltr.io/ Please keep in mind that both Translator and mediKanren are designed for researchers and for biomedical research, not for patient care or treatment recommendations.
- smeej 2y ago
  It's surprisingly reassuring to realize it wasn't actually my fault that I couldn't figure it out. Thanks for explaining! I'll look into Translator and see if I can get anywhere that way.
sn9 2y ago
Have you read their document [0] on how to navigate the process? [0] https://bertrand.might.net/articles/algorithm-for-precision-medicine/ https://bertrand.might.net/articles/algorithm-for-precision-...

isoprophlex 2y ago

That was an incredible read, a real-life Dr. House episode. And very exciting technology, too. Thanks for sharing.

riffraff 2y ago

fantastic talk, thanks for sharing.

dang 2y ago

Discussed a bit on HN: The Algorithm for Precision Medicine, talk by Matthew Might [video] - https://news.ycombinator.com/item?id=37814662 https://news.ycombinator.com/item?id=37814662 - Oct 2023 (14 comments) The Algorithm for Precision Medicine - https://news.ycombinator.com/item?id=29483549 https://news.ycombinator.com/item?id=29483549 - Dec 2021 (1 comment)

fhe 2y ago
thank you for posting this... and this is a common enough pattern (looking up previous HN threads on the same or related article) that I wonder if it's worthwhile to automate this.