AI Across the R&D Value Chain: Clinical Development

That mindset is what’s helping Sanofi reimagine the clinical trial process. AI brings together massive amounts of data, from previous studies, patient records, and even real-world trends, and helps us simulate, predict, and make better decisions at every stage.

For example, imagine you’re planning a road trip. Normally, you’d need to map out the route, and then incorporate real-time information like traffic, weather, and road closures to make sure your journey is smooth. AI is like having a super-smart co-pilot who already knows the best route, where to stop for gas, and how long the trip will really take, because it’s learned from thousands of other travelers.

One major benefit is how we can use data in clinical development to reduce the burden on patients. By using data previously generated in our clinical trials, we can minimize the number of patients or participants required to run a study, especially by reducing the number of patients in the control group (the people who receive the standard treatment or a placebo). In pediatrics, this is especially powerful as “we can use adult data to extrapolate how a drug might work in children. This means fewer children need to be entered into our studies, ultimately reducing the time it takes to make the medicine available for children,” explains Corsico.

Time is another big game-changer. What once took days, now takes minutes. “AI tools can scan, summarize, and organize information almost instantly, helping researchers focus on insights,” says Corsico.

And it doesn’t stop there. AI can also help tailor clinical trial awareness materials—like flyers or videos – to specific patient groups, ensuring they’re inclusive and culturally relevant. “These tools are reconnecting clinical research to patient care,” says Bascles. “AI helps us design studies that truly reflect who patients are and what they need.”

Speed isn’t just about being faster. It’s about doing it smarter. AI is already helping Sanofi find better trial locations, recruit the right patients faster, and cut down on delays. In some cases, it’s helping us skip unnecessary trial phases altogether by simulating outcomes in virtual patients. Digital patient twins (or virtual patients) can help us better understand a disease’s complexities, improve the probability of success in the clinic, optimize clinical trials and create potential best-in-class medicines. We believe AI will further enhance the accuracy and applicability of these digital patient twins.

“By using data to model the efficacy and safety of a medicine in clinical development, we might be able to accelerate the development process,” says Corsico. Models also may enable us to end development sooner if it suggests the drug will not work, sparing patients and allowing us to focus on projects more likely to succeed.  

Clinical development is full of tough decisions. Where to run a trial? Which patients to include? What’s the best design for the study? AI offers a way to challenge our biases and help us make the best decisions by rapidly integrating several facets, such as time, cost, probability of success, and the competitive environment. It also allows faster decision making by providing an unbiased assessment through data review and modelling that was not possible a decade ago.

When developing vaccines, one of the first challenges in clinical trials is figuring out how long it will take to recruit participants. This process, called feasibility, helps our researchers estimate realistic timelines, but it involves a lot of uncertainty. For example, to find patients with specific medical conditions or in certain age groups, we must predict the number of eligible participants in different locations. AI can analyze historical data and disease patterns for more accurate predictions, making recruitment more efficient and reliable.

“With vaccines, it’s crucial to predict where and how quickly a disease might spread, and how severe it could be,” explains Mornet. “AI transforms this process by giving us a clearer picture of key factors, such as where we can recruit and how many patients will be available, versus making educated guesses.”

Of course, AI isn’t perfect. It doesn’t predict the future. It works with the data it’s given, and sometimes, that data has gaps or hidden biases. And like any new technology, not all AI tools are created equal. Choosing the right ones, and knowing when to trust them, can be a challenge.

“We always ask: what don’t we know? What data is missing?” says Mornet. “That’s how we make sure our decisions stay grounded.”

Despite challenges present in the adoption of any new technology in its infancy, the future is exciting. Our teams imagine a world where starting a trial takes weeks, not months—and where AI securely allows us to unlock access to previously inaccessible medical insights and siloed, proprietary and personalized datasets while preserving patient privacy. 

"My 'impossible dream' is that AI will speed clinical processes—one month for protocol approval, one month to launch a study, and 3–6 months for patient recruitment, instead of the current 12–18 months,” added Bascles. “The faster we can collect data, the quicker we can deliver medicines to patients."

“As access to data expands and computer power and speed evolve, new insights will be generated faster and with greater precision. This will allow us to improve how we design our trials, identifying patients more likely to respond to new medicines,” explained Corsico. “I have been impressed by the passion and willingness of everyone at Sanofi to explore how data and technology can be used to make a difference for patients.”

This was part two of our four-part series exploring how Sanofi is using AI to transform the way we bring medicines to life. In case you missed it, read Part 1 on AI in drug discovery here.

Stay tuned for Part 3, where we’ll explore how AI is reinventing manufacturing and supply chains, ensuring that once a treatment is developed, it reaches the people who need it, without delay.