When Genentech licensed experimental drug vixarelimab from Kiniksa Pharmaceuticals, its scientists thought they had a future lung medicine in their hands. Artificial intelligence came up with another plan.
It was an AI platform that led Genentech–a U.S.-based biotechnology unit of Switzerland’s Roche–to discover that the drug candidate it had licensed for its potential to target a lung condition could also work to treat inflammatory bowel disease.
“It’s like, in some way, searching for a needle in a haystack,” said Aviv Regev, Genentech’s executive vice president for research and early development. “We did it based on data and algorithms. We didn’t do it based on ‘now we go back to the lab and we do large-scale experiments and we use these experiments in order to understand the biology.'”
Executives and industry experts caution that AI adoption is still at an early stage and that it will take time, possibly years, before its potential is achieved.
Still, some big pharmaceutical companies are ramping up the use of AI to discover and develop drugs. Their hope is that the technology will help them accelerate and optimize a long and expensive process that often fails to deliver results.
A tally by Boston Consulting Group found at least 67 clinical trials that were under way in 2023 evaluating experimental drugs and vaccines in which AI either discovered, designed or repurposed molecules or target diseases. That compares with 40 clinical trials in 2022 and 27 in 2021, according to the Boston Consulting Group.
British drug giant AstraZeneca’s use of AI for drug discovery across its portfolio shortened the time it took to design molecules from years to months and sometimes even weeks, the company’s chief data scientist, Jim Weatherall, said.
London-based GSK said it sees potential for AI uses across its entire pipeline, from selecting and prioritizing which genetic targets new drugs should aim for to understanding which patients at which point of their disease will best respond to drugs.
“We now routinely integrate millions of data points from laboratory experiments, clinical studies and real-world data to better understand the drivers of disease and then use this knowledge to precisely design and target treatments,” the company’s senior director of machine learning and AI, Patrick Schwab, said.
To fulfill AI’s promise, firms are juggling the need to ensure they have the right technical capabilities, people who understand them and data to underpin the whole effort.
“There is a data problem, a people problem and a bit of a strategy problem,” said Thomas Clozel, chief executive and co-founder of Owkin, a Paris-based company that applies AI to medical research.
Access to data and to skills specific to AI are among the key hurdles pharma companies are facing to unlock the technology’s potential. Clozel pointed to a focus on short-term results among pharma companies and difficulties attracting and retaining data scientists as factors hindering the industry’s embrace of AI.
“Technology is not the primary bottleneck preventing faster progress; it’s largely operational and cultural challenges,” Novartis’s head of AI and computational sciences, Bulent Kiziltan, said. “The hierarchical and linear operating culture in pharma often slows innovation.”
Several drugmakers entered into partnerships with AI specialists in a bid to leverage the technology in their operations. Eli Lilly, Moderna and Sanofi partnered with ChatGPT maker OpenAI, AstraZeneca and Sanofi joined forces with Owkin, and Novartis and Eli Lilly teamed up with Isomorphic Labs, an Alphabet-owned company founded and led by Nobel Prize winner Demis Hassabis, among other collaborations.
Source: Livemint
