Applied Clinical Trials at the 2026 DIA Global Annual Meeting: Conversations on Data, Outsourcing, and the Patient Voice

The 2026 DIA Global Annual Meeting brought thousands of clinical research professionals to Philadelphia, PA this June, and Applied Clinical Trials was on the ground for it.

Over two days, ACT moderated a session on outsourcing strategy and sat down with seven experts spanning regulatory policy, clinical technology, site operations, and patient advocacy. What emerged across those conversations was less a single theme than a set of overlapping pressures—on data quality, on operational agility, and on who is actually accountable when things fall through the cracks.

Here is a look at who we spoke with and what they had to say.

The view from the top

DIA Global president and CEO Marwan Fathallah highlighted the meeting's main purpose with a broad read of the forces converging in Philadelphia.

"Clearly the world is evolving—because of the innovation that's happening around the world, because of artificial intelligence (AI), because of exciting science coming forward at a very fast pace to save lives and improve quality of life," he said. "There are also trends in regulatory policy, and I'd be remiss not to mention the instabilities we see around the world."

On leadership, Fathallah added: "It takes a village to get things moving, and leadership is key. Without leadership and a clear vision and purpose, it's really hard to navigate."

Looking ahead, he singled out three areas generating the most excitement—obesity therapies, healthcare AI, and cell and gene therapy—describing the pace of progress in each as something the field could not have anticipated even a few years ago.

Rethinking what "real-time" actually means

Kevin Bugin, head of global regulatory policy and intelligence at Amgen and executive sponsor of TransCelerate's Embedded Pragmatic Trials initiative, offered one of the meeting's sharper reframes on where regulatory expectations are actually headed.

"They're not focused on what I'd call real-time clinical trial data streaming to the FDA," he said. "They're really focused on signals—and those signals are intended to inform critical decisions. If anything, they're focused on real-time evidence generation for regulatory-grade decision making."

On data quality, Bugin pushed back on a deeply ingrained industry habit.

"The modus operandi for many, many years has been to generate more data, then do a lot of data cleanup and data management to ensure that data will be high quality. What we're hearing from regulators is: no, you need to focus on the right data at the right time and in the right standard. That quality comes from quality-by-design thinking, not something you can build in after the study is over."

Protocol digitization as the upstream enabler

Angie Maurer, VP of AI-enabled clinical development at Medable, traced the protocol digitization bottleneck to three structural barriers that have kept automation out of reach until now: the absence of a standard data model, the inherent ambiguity of clinical language, and zero error tolerance requirements that locked sponsors into manual processes.

"Today there's a convergence of three things happening at once," she said. "LLMs can now decipher the nuanced clinical language, industry standards like USDM give structured output somewhere to go, and we now have the ability to operate within a validated environment."

The implications for amendment management are significant. "When you have a protocol that is structured data, it gives you a clear dependency map—a clear path for where that data needs to go, and a more automated way to get it there," Maurer said. "When you have a change in a visit window, that's no longer just a text edit. That visit window change becomes a data object that has relationships to other data objects downstream."

For Maurer, this is not just a startup efficiency gain—it is what makes the FDA's continuous oversight vision operationally achievable.

Mixed outsourcing as strategic tool

Applied Clinical Trials moderated a session on hybrid functional-service provider/ full-service outsourcing (FSP/FSO) outsourcing models and sat down with both panelists after.

Nick Scott, head of strategic resourcing and performance optimization at Biogen, described the transition as a strategic evolution: "The mixed model is really enabling us to make more agile decisions, apply resources in a more agile way, and leverage resources in areas where we don't have a core competency or don't see ourselves needing one."

Samantha Hadfield, VP of operational delivery at PPD FSP Solutions, Thermo Fisher Scientific, added a geographic lens: for large pharma companies, the mixed model often means dedicated FSP FTEs in core markets paired with unitized FSO arrangements where volume doesn't support the full model.

On implementation, she was direct about what makes or breaks the journey: "The devil is in the detail. You shouldn't underestimate the level of accountability you need to assign, or the level of partnership across workstreams to implement a model like this. And that journey has to be a very deliberate one."

Scott echoed the partnership-first framing: "The number one takeaway for any partnership is that it's a partnership. When challenges arise, you come to the table together, identify the cause, fix it together, and don't point fingers."

The patient voice and federated AI

Stacy Hurt, chief patient officer at Parexel, brought a patient data angle to the meeting's AI conversations.

"When we talk about meeting patients where they are, federated AI helps us meet the data where it is," she said. "Instead of centralizing the data, we can aggregate it across many institutions, community sites, and others—and really represent patient needs in a very broad way without sacrificing privacy."

Her more pointed message, though, was about when the patient voice enters the process.

"The patient voice gets lost in clinical development from the beginning—because that's where it should be," Hurt said. "By the time a protocol is finalized, it's often too late."

On accountability, she added: "I don't think anyone should assume that someone is looking after patient needs, because oftentimes that is getting lost across clinical operations, data, pharmacovigilance."

Site-level data and the integration problem

Jonathan Andrus, co-CEO of CRIO, addressed the fragmentation between site-level systems and sponsor oversight—a problem the field has been circling for years. His framing was straightforward: sponsors have historically viewed site systems through the lens of their own submission needs, while sites have built processes around their own operational realities. Closing that gap, he argued, requires giving sites a central eSource template while still preserving their ability to configure for local needs.

"It's the combination of providing a central eSource template with the ability for sites to still customize and create local configurations that creates a powerful solution—one that meets both parties where they are,” he said.

The session Andrus moderated on this topic drew a standing-room crowd. The takeaway he carried out was simple: "Don't get so hung up on perfection. Focus on making needed changes and taking the steps necessary to continue moving forward. Nothing is going to be perfect, but the only way to continue moving the bar forward is to start doing things."

Decentralized trials: Flexibility with caveats

Following a session she moderated, Joan Chambers, independent senior consultant at Tufts CSDD, offered a grounded read on where decentralized clinical trial (DCT) models actually stand.

"DCT really offers patients the flexibility, accessibility, and convenience of participating in clinical trials," she said, describing telehealth visits, mobile apps, in-home nursing, and mobile vans as modalities that can be deployed selectively depending on the study.

The value, she argued, is in expanding access—particularly for rural populations facing geographic barriers.

On the operational side, she was candid about site burden: "Sites are managing a multitude of different studies for different sponsors, and based on that they're already using different technologies. Now you're adding an extra layer—this DCT component—and that adds a level of complexity."

Her takeaway was measured: "DCT is not going to be one size fits all. It really has to be fit for purpose. The data is going to tell the story. But right now we're still learning, still adapting."