Dr. David Fajgenbaum

It happens sort of randomly, and there isn't like a system that exists.

There's never been an entity, whether it's in the government or within industry, that's just been focused on saying, okay, among the 4,000 drugs that are approved, that are available, what are all the other diseases they can treat?

and because it hasn't been done systematically, I could discover a drug like serolimus could save my life, and drugs can be found like tocilizumab could be used in all these different ways.

So there's all these drugs out there, and there's just never been an entity to say, like, we're going to match in everything we can.

We've advanced 14 drugs, 14, for diseases they weren't intended for.

14 drugs.

Yeah, 14.

Isn't that incredible?

And that's in the last just 11 years, since serolimus, of course, being one of them.

And that's primarily from my lab at Penn, where we're focused on just a few rare cancers and inflammatory diseases.

And so about three years ago, I'm sitting here thinking, okay,

In our small lab at UPenn, we're advancing, at that stage, it had been over 10 drugs for diseases they weren't intended for.

It's like, how many more are there out there?

And what was so important that this came up about three or four years ago is that one of my really close friends from medical school, Grant Mitchell, was really pioneering the use of artificial intelligence to use medical record data to try to figure out

subtypes of patients that could benefit from one drug or another drug.

And we came together and we said, what if we could utilize artificial intelligence to scan across every drug and every disease?

Like my lab works on a few diseases and a few drugs.

What if we could actually scale what we're doing for these few, but to all?

Like what if we could use AI to look across all of the world's knowledge to come up with

how likely every drug, all 4,000 of them, are to treat every disease, all 18,000 diseases.