Loading video...

Video Failed to Load

Go Home

In what order will AI solve the three big problems in drug development? Ben Liu, co-founder of Formation Bio: "There are actually three consequential predictions: drug design and discovery, tox prediction, and clinical efficacy. We believe they will all one day happen, but the winners will be disproportionately right...

17,344 views • 2 months ago •via X (Twitter)

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

$ABCL CEO Carl Hansen explains the net negative overall value creation in biotech and how AbCellera plans to beat it: “I really think it can be beat because there’s a lot of mistakes that get made in where you allocate capital.” “If you’re a biotech and you start a company on a drug, and you go public and have a drug, and now the data is looking not that great, you just keep running that drug. … Because if you quit that drug, the company is done.” “So, because you’re so concentrated on one thing, you end up having blinders. It’s like a cult of a molecule. And even a real, practical, capital access problem that makes you continue to move things forward” ______________________________________ “Another problem is that many of the drugs that get started, get started by people who don’t understand the competitive or commercial landscape and have no intent of ever actually manufacturing it. … Because they know someone else is going to buy it. So then their framework is not ‘Am I making a drug that’s going to make a difference for people?’ The framework is ‘Can I make something that I think someone will take off my hands?’ And that happens a lot.” “There are many acquisitions where a big pharma company spends $10B, $20B, and it’s a complete bust. So really no value is generated for patients but someone was able to get a good exit on that because they anticipated what someone would want. And a lot of decision making gets done like that”. _______________________________________ “I’m not saying that we have this all solved. But what we’re trying to do is create a framework where we are able to take many bets so that we don’t fall in love with anything, or we don’t have to fall in love with anything. … And then always hold up very explicitly what we know and what we don’t know along the dimensions that matter. The dimensions that matter are - Will it work? That’s a science. - When it gets where will people care? - Is it differentiated? - Does it solve a big problem? In order to get it there, is there a path with our resources and expertise that would allow us to see it through?” “When you start to hold things up like that, you start to see where the mistakes are, where the uncertainties are, and you can be more rational in deciding ‘We’re going to run it to here because we get to flip the card and if the card goes the wrong way we’re going to kill it because we’ve got something else behind it.’” “I do think that if done right, and if you can pick the right opportunities, the success rate can be well higher than an order of magnitude better. … And there are examples of this… companies I’ve mentioned, you know, Regeneron, their success rate is at least 10x the industry in bringing programs forward and getting them approved. … So it’s… you can do it better that way”

Jack Prescott

31,665 views • 1 month ago

I am currently fundraising to build the pharma company of the future. Please email me at [email protected] if you are interested (angel checks + VC welcome). Advances in AI + robotics now enable a ‘single’ human operator to carry out the entire scientific discovery process for a new drug, while costing orders of magnitude less than usual. As proof, I developed two new drugs by myself, with zero outside investment, and assistance only from frontier AI models and liquid-handling robotics. PAC-832 is the world’s first selective GalR1 antagonist for Alzheimer’s disease. PAC-3310 is a new selective M4 agonist for schizophrenia, improving on the selectivity profile of the breakthrough drug Cobenfry (which spearheaded the $14B acquisition of its developer Karuna Therapeutics). This level of efficiency is unprecedented and will enable me to generate >40x more clinical-stage drugs (per dollar spent) than a typical pharma company. Pfizer currently has the most clinical-stage drugs at 137. I will be able to reach 10 clinical-stage drugs within 2 years, then surpass Pfizer within the next 5 years. This plan is contingent on me optimizing one more set of studies known as ‘IND-enabling studies,’ which are required by international regulations to be done prior to clinical trials. This involves me (1) building out my own GMP drug manufacturing facility in SF, and (2) optimizing GLP toxicity studies, which I will be doing in China for cost reasons. Both these things are critical to keep overall costs low and require significant resources, which is why I need to raise money, but are eminently doable. I expect them to take 1-2 years in total. After that, my optimized drug synthesis-to-clinical testing pipeline will be done, and I will start turning the flywheel. The long-term, hardest, yet most important task will be optimizing the clinical trials themselves. In short, I intend to accomplish this by launching the most comprehensive, worldwide site search campaign that I can muster. Once the clinical trials are optimized, then I can finally close the loop on the make -> test -> iterate cycle whose speed dictates progress in drug discovery. The 1950-60s are commonly referred to as the ‘golden age of drug discovery’ due to the large number of transformative medicines that emerged from that era. The key to their success was their tight clinical feedback loop - new drugs were synthesized and rapidly tested in the clinic within 1 year. Today, this process takes 10x as long. Enabled by new technologies like AI and robotics, I strongly believe it is possible to bring back the old level of speed and efficiency. A new golden age of drug discovery is on the horizon, and Pace Pharmaceuticals is its herald.

Douglas Yao

180,635 views • 6 days ago