When we think of artificial intelligence (AI), data science and the potential real-world impact of these rapidly developing fields, we tend to look to the future—particularly when it comes to health.

In life sciences, AI is already having a significant impact on our ability to bring new medicines to patients, from accelerating drug discovery to supporting the early detection of disease, improving clinical research and shaping our understanding of complex biology.

At AstraZeneca, an early adopter of AI and data science for healthcare, these tools are helping scientists in their quest to find life-changing new treatments for some of the world's most challenging diseases, in order to transform patient outcomes.

Jim Weatherall is the Chief Data Scientist, Biopharmaceuticals R&D at AstraZeneca. For the past five years, working closely with partners across the organization, he has played a key role in integrating data science and AI into drug discovery and development—from answering specific scientific questions to improving the ways of working for chemists, biologists, and clinicians who are striving to solve them.

"Biology is still an unsolved problem," Weatherall says, "It's characterized by a lot of uncertainty and a lot of variety—variety between individuals, between cells in the body, variety between what the proteins in those cells do and how they talk to each other.

"There's still so much we don't understand. However, the good news is that we have extensive historical data from experiments and clinical trials. We're also continuously generating new data, more than ever before, and can use AI to turn it into value for patients at unprecedented speed and scale, in order to deliver much needed biomedical insights."

One of the key ways in which AI has transformed AstraZeneca's research is in the discovery and design of drugs. Historically, medicines have been discovered and designed manually by scientists in the laboratory—this needs to be rigorous and deliberate, but can often be a slow and painstaking process, contributing to the lengthy development process of new medicines.

Now, AstraZeneca chemists are armed with AI tools inspired by "reinforcement learning." This technique is, in part, inspired and borrowed from its success in other tasks—such as applications where computers can exceed human levels of performance in winning at games such as chess and Go.

"In reinforcement learning, you're asking AI to maximize a score, to find as many different outcomes as possible and choose the best one—which is something that can be applied just as well to the design of a molecule that might one day become a new medicine," Weatherall explains.

It means that instead of synthesizing hundreds of molecules by hand, machine learning tools can quickly sift through thousands of possibilities, highlighting the most promising candidates for further study.

It also allows researchers to take a more "zoomed-out perspective," Weatherall says, "to help us make better scientific decisions, so that we can have a higher probability of success, and the potential for medicines to reach patients faster."

Going forward, AstraZeneca researchers are applying that same approach to different classes of medicines, such as antibodies, which are often used to treat diseases.

Antibodies' complexities require thinking around biological processes at the molecular level, Weatherall explains: "In the past, that has been a laborious process, but now we have the ability to screen up to hundreds of millions of antibody sequences per experiment. This unlocks our ability to train Machine Learning models, meaning in the future we can use AI to design antibodies to target a particular disease at speed. ."

And wherever greater volumes of data are being generated, AI is beneficial in ensuring that researchers "don't get lost in a sea of data," he adds. "It can help us find patterns and understand novel ways to target disease."

AstraZeneca takes a creative approach towards AI integration, not to replace all human decision making, but to streamline the process to enable faster and safer drug development—with the aim to improve the lives of patients.

"When we can combine AI with new experimental paradigms, we start to turn parts of the drug development process into something hugely accelerated compared to what it was before. In fact, we're now at a point where more than 85 percent of AstraZeneca's small molecule pipeline, a type of medicine which is widely used, has had its development supported by AI tools." he says.

AstraZeneca is also making strides in its use of AI within clinical trials, one example being in respiratory medicine.

Typically, medical professionals manually listen back to trial recordings to count the number of times a patient coughed—but through training an AI program to recognize and classify different types of audio, in the future it will be possible to not only free up valuable human time, but to achieve more accurate results.

Recognizing how quickly this field is evolving, and far from exploring all the possibilities of AI alone, the company operates in an interconnected, interdisciplinary ecosystem, working with both commercial partners and academic groups seeking to push the cutting edge of research and meet the new challenges posed.

In 2020, Weatherall cofounded the Cambridge Centre for AI and Medicine. Working alongside another pharmaceutical peer to maximize inputs and insights, he worked with a number of departments at the University of Cambridge on high-risk research projects.

"Partnerships are absolutely critical, and long gone are the days where we would try and do all this in isolation," he says. "By fostering a culture of intellectual curiosesity and collaboration, we want to empower our data scientists and partners to push the boundaries of what's possible in healthcare."

To deliver on the promise of AI and data science, ethical considerations such as data privacy, transparency, and algorithmic bias are paramount.

"In life sciences, patient safety and public health are our highest priority," he explains. "At AstraZeneca, ethical and responsible use of data and AI is at the heart of everything we do—by ensuring appropriate standards are in place, we can be confident in our results."

"AI is already transforming the way we discover and design medicines," says Weatherall. “The use is no longer an 'add on' or 'nice to have,' but an essential and integrated part of the science we deliver. If you want to work on really hard problems with high stakes solutions, life sciences is a fabulous place to be—you have the potential to help more people live better, healthier, longer lives.”

This article has been initiated and funded by AstraZeneca. Z4-71295, January 2025.