Finding a potential blockbuster medicine typically takes years of extensive analysis in the laboratory, with teams of researchers methodically sifting through data and test results to unearth a promising candidate. But when Japan’s Takeda Pharmaceutical Co. bought an experimental psoriasis drug for US$4 billion from a Boston start-up in February, it gained a compound selected in only six months by using artificial intelligence.
In the coming months, the drug—picked out of thousands of potential molecules through AI and machine learning algorithms—will progress to the final stages of clinical trials. If successful, it could become one of the first therapies discovered with the help of AI. Analysts at Jefferies estimate it could generate as much as 500 billion yen (US$3.7 billion) in annual sales.
The Japanese drugmaker’s push comes at a time when pharma companies globally are embracing AI by striking deals with computer-savvy startups and adding more data scientists of their own. Their hope is to cut costs and speed up time to market. Morgan Stanley estimates that over the next decade, the use of AI in early-stage drug development could translate into an additional 50 novel therapies worth more than US$50 billion in sales.
Research firm Deep Pharma Intelligence estimates that investments in AI-driven drug discovery companies have tripled over the past four years, reaching US$24.6 billion in 2022. In January of last year, Sanofi agreed to pay UK-based Exscientia Plc US$100 million upfront, plus the potential for up to US$5.2 billion in milestone payments to research novel medicines and develop up to 15 candidates in oncology and immunology with the use of AI systems.
Bayer, Roche Holding and Takeda are among the companies working with Recursion Pharmaceuticals Inc. in Salt Lake City to explore drug discovery using machine learning. Meanwhile, AstraZeneca Plc formed a partnership with BenevolentAI in the UK and Illumina Inc. in San Diego for similar efforts.
See also: Testing QA New Section BDC Feature Winner 1
“When biopharma companies successfully apply AI in R&D, there can be a significant impact,” says Alex Devereson, a partner at McKinsey & Co. who advises drugmakers on digital processes and analytics. “In five years, we expect these approaches to become more structurally embedded in pharma R&D processes and lead to more impact at scale.”
While AI can help, scientists still have to do lots of traditional legwork after the molecule is chosen. The Takeda compound went on to require years more of human clinical trials and other testing. And AI has other limitations. For instance, it can’t predict complex biological properties, such as the efficacy and side effects of compounds.
Still, using technology to identify the next blockbuster therapies can help eliminate some of the guesswork that typically requires hundreds of lab experiments—often spread over many years—to identify promising molecules.
See also: Unpublished article shouldnt be accessible testing
Big Pharma became more serious about investing in AI and machine learning, or ML, after 2018, when Google parent Alphabet Inc.’s DeepMind unit used an AI programme called AlphaFold to beat a biologist at predicting the shape of proteins, the basic building block of diseases. Figuring out the shape of proteins, one of the most vexing problems in biology, helps drug hunters narrow down molecules that might interact with them and identify medicines to attack disease.
Bringing a new drug to market has traditionally cost almost US$3 billion, and about 90% of experimental medicines fail. So technology that speeds up the process can be a big driver of profits. Determining the 3D structure of a protein now takes seconds using AlphaFold, as opposed to many months or years, according to Eric Topol, founder and director of Scripps Research Translational Institute in California, cited on DeepMind’s website.
The growth in AI adoption by pharma companies was accelerated by the Covid-19 pandemic, as the industry rushed to develop weapons to fight an unknown virus. During the pandemic, Pfizer Inc. turned to AI to develop the Covid vaccine Comirnaty, for which it partnered with BioNTech SE. It also expanded a partnership with Shenzhen, China-based AI drug discoverer XtalPi Inc. to speed up the chemical formulation of the Covid pill Paxlovid. Both were approved by the US Food and Drug Administration in under two years—much faster than the 10 years it usually takes most drugs to get to market. The speed was also helped by regulators rushing to get weapons against Covid to the public.
Takeda’s experimental drug, purchased from Boston-based Nimbus Therapeutics LLC, would be one of the world’s few oral treatments for psoriasis, a skin condition that afflicts 125 million people worldwide. It also has the potential to treat other conditions, such as Crohn’s disease, an inflammatory bowel disorder. The drug, currently called TAK-279, has already successfully gone through the first two stages of human trials. Algorithms picked the present molecule in about a quarter of the two years a traditional approach would take, says Jeb Keiper, chief executive officer at Nimbus.
Scientists testing the chemicals in beakers would need to test many molecules—“an impossible number,” says he. Instead of coming up with tens of thousands of compounds to figure out, computers suggest testing 10 compounds in a lab, and then getting feedback from the lab results. The machines learn from those results to make a better prediction to provide the next hundred candidates for testing and ultimately filter to one molecule, Keiper says.
These days, more than 500 quantitative scientists and tech experts in Takeda R&D centres from Boston to San Diego to Shonan in Japan spend their days crunching data to find, develop and manufacture breakthrough medicines. The drugmaker uses AI and ML to identify the best molecules for targeting proteins and to understand the characteristics of diseases and how they vary in different patient populations. It works with the Massachusetts Institute of Technology and several AI startups.
“Any technology that unlocks cutting-edge skills for our employees, reduces manual work, takes the friction out of the system and frees up time for greater scientific insight and discovery is vital,” says Anne Heatherington, head of Takeda’s data science institute.
To stay ahead of the latest tech trends, click here for DigitalEdge Section
Takeda’s larger rivals are also tapping AI. Pfizer expects a partnership with DeepMind’s AlphaFold to help the company design and validate highly effective therapeutic targets that were previously unknown, says Lidia Fonseca, chief digital and technology officer at Pfizer. “We used powerful supercomputing capabilities with AI and machine learning models to reduce our overall computational times by 80% to 90%, and that actually helped fast-track Paxlovid,” Fonseca says.
Around the world, several potential drugs that were identified by startups using AI are already in human trials. They include five from Recursion Pharmaceuticals Inc. for rare diseases and oncology and three from Exscientia for illnesses such as cancer and obsessive-compulsive disorder. Hong Kong-based Insilico Medicine has a candidate in midstage human trials to treat the most common form of pulmonary fibrosis.
UK-based GSK Plc has more than 160 experts dedicated to AI and ML to support its R&D and manufacturing. It also generates data to build and feed the company’s own machine learning models so every scientist can ultimately benefit from data the company has produced in the past, says Kim Branson, who’s been the head of AI at GSK since 2019.
China is also looking toward AI to enhance its drugmakers’ global competitiveness. XTalpi is partly funded by Chinese tech giant Tencent Holdings Ltd., while Baidu Inc. CEO Robin Li founded an AI-driven drug discovery company called BioMap.
GSK’s Branson says that while AI is really good at piecing together data from disparate sources, things get tricky when it’s used for complex systems. To ensure safety, lab experiments are often necessary, he says.
Also, the data used to create algorithms can contain bias that can be reflected in the clinical recommendations they generate, researchers at Stanford University wrote in a study published in the New England Journal of Medicine in 2018. Algorithms can also skew results, depending on who develops them, the researchers found.
That isn’t stopping the jump in investments. There’s been a surge in venture capitalists requesting evaluations of potential AI drug discovery companies over the past five years or so, says Russ Altman, a professor of bioengineering at Stanford, who’s conducted due diligence of biotech startups for VCs for decades. “It went from zero to a hundred,” Altman says. “I hadn’t done any due diligence on AI drug companies in 30 years. And now I’ve done six to 10.”
