The world's biggest pharmaceutical companies are engaged in a talent tug of war with Silicon Valley as they seek to harness the power of artificial intelligence to accelerate the discovery and development of new medicines.
Companies like AstraZeneca PLC, Novartis AG and GlaxoSmithKline PLC are persuading data scientists to fill newly created roles traditionally associated with technology companies, in a bid to bring outside-the-box thinking and a more nimble execution to an industry that grapples with long development timelines and huge costs.
AI has been deployed in drug discovery, clinical trials and manufacturing. Richard Dearden, a former NASA scientist, joined Cambridge, England-based AstraZeneca in 2019 to help a team of data scientists transform and accelerate their clinical trials, in order to get drugs to market faster. Dearden is banking that experience gained from programming the Mars Rover to move and collect data millions of miles from Earth, as well as years spent in the oil and gas industry, will improve efficiency in supply chains and reduce the burden of trials on patients.
"I started in AI just after the '80s, when we had our previous golden time in the sun — so I've been through the whole 30 years of AI being a dirty word in industry, and it's lovely to be back in the sunlit highlands of people's minds again," Dearden told S&P Global Market Intelligence. Still, he acknowledged that AI alone will not solve the industry's problems and human expertise remains crucial. "There are very few situations where you gain insights without understanding the problem as well," he said.
Novartis' clinical trial command center at its Basel headquarters.
While big pharma is not alone in turning to AI to combat inefficiencies in the supply chain and manufacturing, companies like Novartis, the fifth-largest drug company in the world by market cap, have harnessed AI to amplify the application of science. With a background in consulting and medicine, Vas Narasimhan, who previously headed the group's R&D operations, has moved to align the potential of science with data and digital capabilities since he became CEO in 2018.
A dedicated command center akin to an airport air traffic control tower has been built at its Basel, Switzerland, headquarters. Within this, a wall of screens relays real-time data for every clinical trial Novartis is running anywhere in the world at any one time. Narasimhan has plans to build similar centers to display all of the pharma giant's manufacturing data, and another to illustrate the people and culture at Novartis, which employs over 108,000 staff globally.
Novartis has also created the ACTalya algorithm tool for its sales force, which refreshes overnight and provides its 10,000 reps with insights about how to best interact with their clients — whether to meet in person or email instead, for instance. Narasimhan said it was notably popular among the sales reps in China.
In 2019, the company took its strategy one step further by beginning a multiyear alliance with Microsoft Corp. to bolster AI capabilities across the pharmaceutical company, from research through to commercialization of drugs.
"We feel like overall in our data science, digital ambitions, we're making progress," Narasimhan said in a 2019 interview.
The pharma industry faces some unique challenges, not least the frequency of patent cliffs, meaning that even once a so-called blockbuster drug has been successfully developed and approved, it has a limited number of years of exclusivity before cheaper generics flood the market. Waste is another glaring issue: With only 10% of drugs making it past phase 2 and into the final stage of testing before being approved for use in humans, selecting the right target early on is crucial, said GSK's head of medicinal science and technology, Tony Wood.
GSK turned to Silicon Valley shortly after appointing Hal Barron — formerly of California Life Sciences LLC, an Alphabet Inc.-funded company looking at how technology can impact lifespan — as head of R&D in 2017. The Brentford, England-based drug giant pivoted its research to focus on the immune system, and Barron signed an alliance with consumer genomics company 23andMe Inc. to mine a rich new source of data in its hunt for drugs. "What we're really focused on is choosing the right targets," Wood told reporters at the company's Stevenage research center in November 2019. "That really is about the genetics and functional genomics."
Not everyone is persuaded. Nooman Haque of Silicon Valley Bank told S&P Global Market Intelligence that AI in the pharmaceutical industry has yet to prove itself. He pointed out that a lot of these tools have been around for a couple of decades, but were previously called bioinformatics. "The cheapness of the technology may be new and the speed and processing power, and access to the cloud — but the impact that makes on the overall pipeline at this stage is quite marginal, because it's only impacting sometimes on that discovery phase," he said. "So you're getting to that 1 in 10,000 compound identification quicker, perhaps."
Good old bioinformatics
Still, AI has been successfully used in managing chronic diseases like diabetes for some time, thanks to the development of wearable technology that monitors glucose levels and links up with mobile devices. Now, researchers are hoping to use AI to develop smartwatches capable of detecting early signs of Alzheimer's disease via changes in gait or sleep patterns.
IXICO brain imaging using the LEAP algorithm.
IXICO PLC, a clinical research company spun out of University College London and King's College London, said there is growing demand for its technology, which has provided expertise in the imaging of neurological diseases for over a decade. While its core business is monitoring the efficacy of a drug in clinical trials via medical imaging, the London-based company is also working on digital algorithms around sleep patterns for Parkinson's patients, among other projects.
"As clinical trials become more complex, we're seeing the [use of] medical imaging increasing," IXICO CEO Giulio Cerroni said in an interview. "[Big pharma] very much do come to us because of our expertise in data analytics around neuro-diseases."
But while the rapprochement between the pharmaceutical industry and the tech sector is likely to increase, analysts don't foresee AI replacing human expertise any time soon, even taking into account the recent explosion in computational processing power.
"This issue is aggravated by the dearth of people who have bilingual capacity to understand both machine learning and the underlying applications to life sciences … resulting in many capable hires remaining within the tech sector, where they command a seat at the highest level," SVB Leerink analyst Mani Foroohar said in a March 4 note.
At the S&P Global Market Intelligence Healthcare Breakfast Briefing in New York on Feb. 27, Stephen Diamond, assistant general counsel in the business transactions group of Pfizer Inc.'s legal division, outlined the cultural barriers that need to be bridged.
"You're talking about two industries that don't really understand each other," said Diamond, who noted that his views at the briefing were his own and did not necessarily reflect those of Pfizer or his colleagues at the company. "Tech is move fast and break things, and pharma is the opposite of that: it's take your time, make sure it's safe and then go out. ... Neither one really speaks the other's language."