Drugs discovered by AI: the state of play May 2024

Reading: How successful are AI-discovered drugs in clinical trials? A first analysis and emerging lessons from the Boston Consulting Group, and the commentary in AI Drugs So Far from Derek Lowe (DL) in Science.

The state of play is:

• 67 molecules in 39 clinical trials
• 114 companies in AI drug discovery space

Are those big numbers? The are certainly rising quickly:

That one “launched” drug is BXCL501, for Alzheimer’s related agitation. It seems to have met its Phase III target, and I believe it is going through the process towards becoming a approved drug.

This is all pretty exciting. As the paper notes: “In Phase I trials, AI-discovered molecules are substantially more successful than historic industry averages”.

DL makes a couple of excellent points on all this:

They say that of the 24 therapies that have reported Phase I, 21 were successful [84%]. The industry standard success rate for Phase I is 66% for all indications and 76% for lead indications, so while the AI-based examples here might be getting through Phase I at a higher rate, both the sample size and outright differences are too small, in my opinion, to make that claim.

And digging in, DL notes:

What you will see is that in almost every case, these targets were already known to be implicated in the disease under investigation. In some of these examples, in fact there are several drugs already in the clinic targeting the same proteins […] I do not see how any of them can be classified as "target discovered by AI". I really don't. 

The point of AI drug discovery is speed and variety: searching for drugs no-one has thought of before, delivered either faster or at less cost.  Exciting, yes? The go-to example is taking just 18 months from target discovery to preclinical candidate nomination.