Novartis will partner with Microsoft in a five-year bet that the storied technology company can help it make better sense of the reams of data generated from its laboratory experiments, clinical trials and manufacturing plants. In a partnership announced Tuesday, the two companies said they would work together to develop applications for Microsoft’s artificial intelligence capabilities throughout Novartis’ drug development, focusing first on personalizing treatment for age-related macular degeneration and on improving manufacturing of CAR-T cancer therapies. Novartis and Microsoft’s vision also extends to the drugmaker’s more than 100,000 employees. Through a newly created AI innovation lab, the companies hope to make AI-powered applications widely accessible and usable across Novartis’ organization.
Data and digital feature prominently in Novartis CEO Vas Narasimhan’s push to rebrand the Swiss drugmaker as a “leading medicines company.”
While no financial details were disclosed Tuesday, the multi-year partnership with Microsoft is evidence enough of that commitment. It comes as more pharma companies are placing chief digital officers on their executive committees, and as more ink partnerships to explore the role AI can play in drug development.
AI’s promise might be great, but its impacts in biomedical research are so far modest, hampered by human biology, messy data, and incomplete knowledge of disease pathologies.
Novartis and Microsoft described their agreement as an alliance, one that will benefit the tech company as much as it does the drugmaker.
“It’s really an equal-to-equal partnership,” said Bertrand Bodson, chief digital officer at Novartis, in an interview with BioPharma Dive. “It’s absolutely not a vendor relationship.”
Employees of the two companies will collaborate at Novartis’ headquarters in Basel, Switzerland, at Novartis’ service center in Dublin, Ireland, and at Microsoft’s research lab in the U.K.
For Microsoft, partnering with Novartis gives it opportunities to test its AI technology for precise uses like in CAR-T therapy manufacturing, as well as at scale across a global company.
Novartis estimates it’s collected 2 million patient-years worth of clinical trial data over the past two decades, while its laboratories hold data on 1.5 million chemical compounds.
“The issue isn’t just a problem of the overwhelming volume,” wrote Peter Lee, head of Microsoft Healthcare, in a blog post announcing the partnership.
“Much of the information exists in the form of unstructured data, such as research lab notes, medical journal articles, and clinical trial results, all of which is typically stored in disconnected systems.”
Novartis and Microsoft will try their hand at solving those hurdles in three areas — two specific, and one broad.
In age-related macular degeneration, for example, the companies plan to use AI to scan images of patient eyes, hunting for patterns that could pave the way for more personalized dosing.
“We have thousands and thousands of OCT scans,” said Bodson, referring to an imaging technique called optical coherence tomography. “That’s a beautiful input for AI, because one of the easiest areas for pattern recognition and patient stratification has been imaging.”
Novartis sells Lucentis (ranibizumab) for a type of AMD and, as it happens, hopes to soon win U.S. approval for a new drug for the condition.
Also on the agenda is using AI to improve the manufacturing of CAR-T cells, a time-consuming and resource intensive endeavor that’s given Novartis headaches in commercial roll-out of its drug Kymriah (tisagenlecleucel).
Novartis and Microsoft also plan to apply AI to generative chemistry, aiming to use advanced computing to design better drugs. Most drugmaker-tech partnerships end up on this path, if only because drug discovery is both data heavy and likely to fail.
“Typically we have to sequentially design a lot of molecules to fit the right purpose,” said Bodson. “We’re hoping to speed and enhance that process, so we can generate molecules that will have the highest probability from a chemical point of view to bind in the right way, to have the right structure.”
Teaming up with Microsoft fits well with Narasimhan’s vision for Novartis. Another pillar of the CEO’s plan to build a leading medicines company centers on developing complex drugs like Kymriah or its gene therapy Zolgensma (onasemnogene abeparvovec).
There, however, Narasimhan and Novartis are enmeshed in a regulatory scandal involving the submission of manipulated data to the Food and Drug Administration. An investigation is ongoing, and could result in penalties for the company.