Between 1796 and 1801, physician Edward Jenner performed 23 carefully designed trials in which he inoculated patients with small amounts of his radical new smallpox vaccine. The success of those trials not only originated the world’s first vaccine—it was also one of the earliest-recorded instances of clinical trials, laying a framework that's in many ways still in place today.
Today, the world looks very different from that of the late 18th century: The global population has octupled; many of us live in dense cities; we can travel from Seattle to Seoul in just a few hours. And with new diseases emerging everywhere, all the time, the need for quickly and rigorously tested medical treatments is more urgent than ever. But despite all this change, rolling out a clinical trial today often takes just as long as it once did (if not longer). The manual methodology is largely still bound by limitations of human capability, and certain stubborn inefficiencies in the pipeline can delay getting critical drugs into the hands of patients who desperately need them.
That’s what leading biopharmaceutical solutions company Syneos Health is looking to change. “There are patients waiting for treatments, and we’re taking too long getting to them,” says Baba Shetty, president of technology and data solutions at Syneos Health. “And AI represented an enormous opportunity to leapfrog the way things have traditionally been done.”
Tightening the timeline, from protocols to procedures
“A clinical trial, if you deconstruct it, is made up of an enormous number of decisions,” says Shetty. “In total, you could say there are 1–3 million decisions that have to be made for any given trial—everything from strategy decisions, like which countries should we select, to day-to-day decisions, like how we handle a particular patient’s situation.”
Each of those human-powered decisions, he explains, could be streamlined and improved if infused with the best-possible data and insights—and the team at Syneos Health knew AI was the tool for the job. So together with their partners at Boston Consulting Group (BCG), they developed what Shetty calls a full-stack, AI-ready platform. “We knew without it, very little else would be possible,” he says.
The teams then began having conversations around one key question: Where, with help from this AI platform—and a set of applications that built upon it—could they optimize decision-making across the clinical trials process?
An early stage of a clinical trial involves developing the protocol: essentially, the scientific “blueprint” for the trial. Generating just a first draft of a protocol can take weeks or months as experts compare the current protocol to similar prior trials, reviewing thousands of prior clinical trials to find appropriate comparisons and learnings. It's an arduous workflow that gobbles up valuable energy and resources.
With Syneos Health’s Protocol AI application, scientific experts can leverage the power of generative AI to simultaneously analyze thousands of past clinical trial protocols, and in minutes, identify a set of similar trials for reference purposes. A human editor can then comb through the AI-generated output and determine whether the selected prior protocols are indeed appropriate, then feed that content back into the generative AI model, along with other specified parameters, to produce a draft protocol for the trial at hand.
As a result, a process that once took months to complete can now be done in a matter of days, Shetty says—representing an unprecedented jump start to the clinical trial process.
Syneos Health’s AI-powered applications also address bottlenecks in task-based trial procedures. For example, one of the most important factors in a successful clinical trial is selecting one or more appropriate research sites—namely, hospital systems and other institutional healthcare providers with access to the right resources and patient pools.
“You have 170,000 clinical trial sites around the world,” Shetty describes. “Which are the best 200 options for a particular trial?”
The Syneos Health study planning application allows them to analyze those 170,000 potential trial sites based on specific criteria: How successful trials at those locations have been in the past, how adept they are at recruiting patients, whether there are sufficient pools of diverse patients nearby, and how many competing trials are nearby, for example. Leveraging a combination of structured data and predictive modeling, Syneos Health analysts can narrow an enormous pool of options and return an optimal plan. To compare, manual site vetting might have previously taken weeks of research and analysis and still not have been as accurate.
It’s a potentially game-changing application of modern technology—and yet one that Shetty simultaneously recognizes could instill apprehension among the medical community; especially considering the inherent sensitivity around clinical trials and their findings. “There’s a need to stay vigilant as editors in an AI-enabled world,” he says. “It could be that there are many absolutely correct answers given by the AI, and then the fortieth answer is a little off. And so you must always ask yourself: Are you still being vigilant at that point? Or have you been lulled into complacency?”
For the team at Syneos Health, this notion is always top of mind. Yes, an AI-based system can instantly scrub hundreds of thousands of existing trials, identify the 20 trials that are most comparable, and tell you why. But it’s up to Syneos Health’s team of human experts to evaluate, agree, or disagree with the output.
An effective symbiosis
This kind of nuanced, reciprocal approach—one where human expertise is unlocked by AI, and AI’s awesome analytical power is tempered by human discernment—is reflected in the essence of Syneos Health’s partnership with BCG and BCG X, BCG’s tech build and design team. Where Syneos Health provides institutional knowledge around clinical trials and their substantive data points, BCG brings the expertise needed to integrate that effectively with AI tools.
“There’s a combination of things that need to come together for large-scale AI initiatives to be successful,” says Satyanarayan Chandrashekhar, managing director and partner at BCG X. “Most importantly, it’s about the chemistry between the team and their partners.”
And the result of all this was not just a precedent-shattering platform, but a methodology that could reshape the future of knowledge work. As Shetty puts it: “It puts the collective applied wisdom of not just an organization, but all of scientific knowledge at your fingertips."
This article was produced by WIRED Brand Lab on behalf of Boston Consulting Group.
