The Algorithm for Precision Medicine, talk by Matthew Might [video] | Hacker News Reader

The Algorithm for Precision Medicine, talk by Matthew Might [video]

youtube.com

71 pointsanjayan14 comments

Please note: the talk is 7 hours, 29 minutes, and 55 seconds into the video. Please append "#t=7h29m55s" to the URI.

This talk was presented at DeclMed 2023 (Declarative Programming in Biology and Medicine), a satellite track at ICFP 2023 (International Conference on Functional Programming). The talk is by Matthew Might, director of the Hugh Kaul Precision Medicine Institute at the University of Alabama at Bermingham.

In this talk Matthew Might takes us on the path that lead him from research in static analysis to his current work in bringing precision medicine to fruition.

Matt Might's efforts, with respect to this path, have been discussed in the past here on HN. First, Hunting Down My Son's Killer [1,2], second, an article in The New Yorker [3, 4,5], third, an article in The New York Times [6], and finally, an interview with him from 2016 [7].

[1] https://news.ycombinator.com/item?id=4038113 (11 years ago, 327 comments)

[2] https://news.ycombinator.com/item?id=16773887 (6 years ago, 11 comments)

[3] https://news.ycombinator.com/item?id=8050106 (9 years ago, 109 comments)

[4] https://news.ycombinator.com/item?id=8049699 (9 years ago)

[5] https://news.ycombinator.com/item?id=16483741 (6 years ago)

[6] https://news.ycombinator.com/item?id=17958424 (5 years ago, 20 comments)

[7] https://news.ycombinator.com/item?id=11366150 (7 years ago, 13 comments)

Got very intrigued by the "Biomedical Data Translator" and found this:

https://ncats.nih.gov/translator

... and this paper:

https://ascpt.onlinelibrary.wiley.com/doi/full/10.1111/cts.1...

... hoping it is the same thing.

Have yet to find some usable code or published tool. Did anyone find any?

They also mention mediKanren, apparently built on miniKanren (which is some language for logic programming).

I personally saw a lot of potential for these approaches, and wrote some future outlook stuff as part of my thesis here:

https://bionics.it/posts/linked-data-science

There is some cool new tech coming out here lately, including TypeDB, which is a little like a modern version of Prolog with proper data storage and less cryptic language:

https://typedb.com/

Unfortunately it seems to lack some of the expressive power that makes Prolog so well suited for integrating basically any outside data source on-the-fly :/

Hi, I'm the organizer of the DeclMed workshop. Yes, nearly all the code for the NCATS Biomedical Data Translator is on GitHub, in a variety of project areas. The best place to start is the Translator developer documentation: https://ncatstranslator.github.io/translator-developer-docum...

once you've got an overview, click on the "Developer Guide" tab on the aforementioned page. Please reach out if you have any questions (https://lab.saramsey.org)

I should have said, co-organizer of the workshop (with Will Byrd and Matt Might). Please also check out Tyler Beck's keynote (on Translator) at 5:24:35, and Sierra Moxon's talk on the Biolink semantic layer at 5:07:23, Chris Mungall's keynote on logic programming at 0:25:11, and Daniel Korn's presentation on drug repositioning at 6:58:01.

Hey sramsey. Glad to see the Translator is still going. I imagine that recent LLM mania has been a pretty big deal for the agent teams? Anyway, nice to see you around and say “Hi” to Noel and Christine for me ;)

My personal take on precision medicine is that before we can truly scale it up, we need to properly understand and treat organisms as _systems_. Thus, no precision medicine without systems medicine.

Systems medicine too is a thing, so the thought is not unique as such. But I just feel there is way too little focus on systems level understanding.

I'm sure there are some counter cases to this, like some examples in the video, where there were some very "simple" treatment solution to a problem. But I think this will not always be the case, and probably in many situations one might need to turn more than one knob in the system to make it "just right" again. At the same time, the knowledge based approach he took to help his son of course kind of embedded a lot of systems understanding in the knowledge bases, so it is perhaps not a complete counter case either.

You may also like my talk about precision medicine in cancer https://www.youtube.com/watch?v=xa1iRNlH7-0

Matt Might is an exceptional human, mind-numbingly sharp, incredibly kind, and an astonishingly great communicator. His precision medicine work has a real shot at saving many many lives.

At 7:47:40 he shows "the interface" and says "please go test it out" but I'm struggling to find where it is. Anyone have any luck locating it?

There is zero description in the video notes, so a description here would be helpful for one deciding whether or not to watch it.

Its a talk at a programming languages conference by a programming languages professor who moved into medicine to help diagnose and treat his son with a genetic condition. Its a great talk which simultaneously provides an introduction to biomedical research and knowledge based systems and provides some compelling cases for the benefit of combining the two. It also takes you through a journey that includes starting a company, working with the white house and being recruited to be the director of a university research institute.

Thanks for sharing - incredibly inspirational on a personal and professional level

One of the most outstanding tech talks I've ever seen.

Reply on news.ycombinator.com