An Interview with AI About Its Role in Drug Development
I interviewed a large language model about how AI is being used in drug development, and where it's going in the future - are we one day going to be able to fully simulate drug development on a chip?
Intro
Artificial Intelligence has been all the rage and BioPharma is no exception. As I’ve touched on in the past, how companies are tapping into the new computing power and digital-to-human interface that advancements in AI have recently brought was a key them at this year’s JP Morgan Healthcare Conference.
I’ve somewhat casually begun trying to learn more about AI in our industry, as it is clear to me that this is already becoming table stakes across industries. However, when you tap into a new podcast or pull up a news article about the topic or see a tweet storm from some thought leader in our field about AI in drug development, it’s all seemed a bit like unicorns and rainbows. All of it sounds a little like,
“AI is already revolutionizing everything about drug development and if you don’t hop on the train now you are going to get left behind.”
“In the near future we are just going be able to invent a drug by typing into a computer.”
“We won’t need to run studies on humans anymore because we can just run studies inside of a computer.”
So, to learn about how AI is hopefully changing drug development for the better, I decided to go straight to the source. Nope, I didn’t get an interview with Jensen Huang or Sam Altman, they unfortunately didn’t return my calls. Instead, I decided to talk to an AI itself – specifically Anthropic’s Claude – and interview it about how AI is being used in drug development, where it may go in the future, and what its limitations may be.
The below is a slightly edited down version of my “interview” with some final take home thoughts from me at the end. It is still quite a bit longer than my usual posts, so feel free to skip to the end for my takeaways and skim through Claude’s responses as needed. I hope you enjoy and learn something from this one. I’ll certainly be revisiting this topic in more detail in future posts.
The Interview
Big Pharma Sharma (BPS): I’m a biotech strategist trying to learn more about how companies are utilizing AI tools in drug development. Imagine you are an expert in AI for drug development. What are the main ways AI is currently being utilized in the drug discovery and development process, and what specific capabilities is it providing today?
Claude: AI is being utilized in several key areas across the drug discovery and development process today. One major application is in target identification and validation, where AI is helping analyze massive genomic, proteomic, and other biological datasets to identify novel drug targets and pathways. Machine learning models are being used to integrate various data sources and predict target-disease associations.
AI is also playing an important role in molecule design and lead optimization. Techniques like deep learning, reinforcement learning, generative models, and docking simulations are enabling the virtual screening of billions of compounds and the design of novel molecules with desired properties. AI is accelerating hit identification and lead optimization.
In the preclinical stage, AI tools are proving very useful for predicting properties like ADME, toxicity, drug-target binding affinity and various pharmacokinetic parameters. This is allowing faster prioritization of compounds and reducing failure rates due to poor ADME/Tox profiles.
Additionally, AI and machine learning are being applied to design and analyze clinical trials more efficiently. Models can help identify optimal trial protocols, predict patient responses, and analyze complex efficacy and safety signals in the clinical data.
So, in summary, AI is impacting target discovery, molecule design, preclinical predictions, and clinical trial optimization. It's providing significant capabilities in data mining, predictive modeling, simulation and optimization across many steps of the R&D process.