Read more.Quote:
Halicin can kill some of the most dangerous antibiotic-resistant pathogens.
Printable View
Read more.Quote:
Halicin can kill some of the most dangerous antibiotic-resistant pathogens.
An excellent use of AI, unlike that other fad of creating dinosaurs pffft
Interesting and well-written article - good to hear about tech's reach into important areas of life that can affect us all.
This approach has to beat the current random scouring of caves and the sea-bed by scientists, all in the hope of stumbling across new naturally-occurring anti-biotics (without a lot of success, so far).
To think all that processing which has gone into bitcoin could have figured this years ago.
Fantastic!
Chemprop.Quote:
Originally Posted by hexus
https://github.com/chemprop/chemprop
Just think of all the processing which has gone into this which could have gone into Bitcoin.Quote:
Originally Posted by Percy1983
Interestingly this is probably also a product of all that health data which has been given to Google to help with projects like this. To much whinging and whining.
Nope. This is a relatively simple application of machine learning, no human data involved other than existing reference sets of drugs approved for use and compound/toxicity databases.Quote:
Originally Posted by philehidiot
Absolutely no reference to the large datasets of lab test results to see which other antibiotics are effective against which diseases? I find that unlikely but I'm open to being educated.Quote:
Originally Posted by kalniel
They don't need human data to find out if an antibiotic works against a particular bacterial species. Remember they're not clinically trialing anything here, but they are selecting from pre-existing compounds. The drug they found is being investigated for diabetes for example.Quote:
Originally Posted by philehidiot
And by relatively simple, I don't mean easy or not clever - primary neat bit in my opinion is how they represented molecules to train the neural network - previously neural nets have failed to be all that great at picking out relevant chemical structural features (I know, I tried to do them for protein molecules years ago), but these guys have found a way of representing the information that's useful. Once they've got that the classification bit is fairly normal, but then they've also done good investigative work in the lab trying to work out how the drug they've identified actually creates it's anti-bacterial effect, and therefore discover a novel action which might open the gates to further drug developments even by conventional means.
OK, so how long before that AI decides humans are the highest threat pathogen on the planet?
If you ask the right (wrong) question, you'll get that answer :pQuote:
Originally Posted by Ttaskmaster
https://images.squarespace-cdn.com/c...e=image%2FjpegQuote:
Originally Posted by kalniel
Ah, but can you trust the answer even if you ask the right (wrong) question?Quote:
Originally Posted by kalniel
Quote:
HAL : I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you.
^nice link back to the name of the compound ;)
I used that line once with a boss I didn't get on with. They didn't get the reference :p (not that I was being serious, so probably a good thing)