Dr. David Fajgenbaum

It's actually each drug-disease match.

If you look in the spreadsheet, it's like DFMO for Bachman-Bopp.

lidocaine for breast cancer, sirolimus for Castleman's.

That's what the row is.

The row is actually the match.

And then there's a score next to the match.

And then you might not see breast cancer again until number 10,000.

The second best for breast cancer might be way below the first best, but it's actually the match that gets the score.

How many matches are there?

75 million.

Because there's 4,000 drugs and 18,000 diseases.

So if you multiply 4,000 times 18,500, you get 75 million.

So basically we're asking the algorithm is like simulate how likely each one of these 4,000 drugs would be to treat each one of these 18,000 diseases.

That is a lot of information to sift through.

It's insane.

Yeah.

And so like, yeah, our team of 50, we're working really hard.

So the team of 50, we've now gone through the top 6,000.

And that's like deep review, like reviewing through to understand what data exists and why it might score well.

Out of those top 6,000, we've...