Last week, we brought together leaders from large pharma, biotech, academia, regulatory science, technology, healthcare, and patient advocacy for the inaugural CONVERGE forum in Boston.

The idea behind CONVERGE was relatively simple: create a space where people who don’t necessarily sit at the same table could challenge each other’s thinking about the future of drug development.

We didn’t want another conference where everyone stays within their own discipline. We wanted different perspectives, difficult questions, healthy disagreement, and, hopefully, a few new ideas.

And we certainly got that.

From the opening keynote on breakthrough medicines and affordable access to our closing discussion on AI, data, and the future of clinical development, one theme kept resurfacing: we have extraordinary scientific and technological capabilities, but realizing their full potential requires us to rethink how we work, how we make decisions, and, importantly, how we collaborate.

As I reflect on the discussions, five takeaways stand out.

1. Innovation is only meaningful if it reaches patients

Professor Kenneth Kaitin of Tufts University opened CONVERGE with a fundamental challenge: can we continue developing breakthrough medicines while ensuring that patients can actually access them?

It was a powerful way to set the tone for the day.

We spend enormous energy improving clinical development: making trials faster, designs smarter, and evidence generation more sophisticated. Yet regulatory approval is not the finish line.

In our interactive forum, Grammati Sarri and Nathalie Horowicz-Mehler from Cytel explored the disconnect between evidence generated for regulators and the evidence needed by payers, healthcare providers, and patients.

Stuart Siedman of Chiesi Rare Diseases brought a deeply personal dimension to that discussion, reminding us why patient communities need to be involved from the earliest stages of development, rather than only when recruitment begins.

The discussion also highlighted something important: involving patients earlier isn’t simply about improving recruitment or retention. It’s about understanding what matters to them, designing studies around meaningful outcomes and ensuring the evidence we generate reflects their lived experiences.

For me, this reinforced the need to think about evidence strategies across the entire patient journey. Not just what it takes to demonstrate efficacy, but what it takes to establish value, enable access, and ultimately improve people’s lives.

Scientific progress only becomes meaningful progress when patients benefit from it.

2. The real promise of AI is better decisions, not simply faster processes

AI was, unsurprisingly, a recurring topic throughout the day. But what I found particularly encouraging was how quickly the conversation moved beyond productivity.

In her plenary, Rethinking Clinical Development: Leading the AI-Enabled Future, Justine Rochon of Takeda challenged us to think about AI as an opportunity to improve decisions under uncertainty.

Can we identify promising treatments earlier? Recognize safety signals sooner? Design more precise trials? Make better go/no-go decisions before committing years of development and significant investment?

These questions are arguably more important than how many hours we can save through automation.

The discussions on next-generation trial methodology, including contributions from Li Wang of Cytel, Shu Yang of NC State University, and Manuel Cossio of Cytel, further explored the opportunities created by combining statistical innovation, external information, and AI-enabled approaches.

Yet there was also an important reality check throughout the day. While AI is already delivering efficiencies in specific workflows, its ability to fundamentally transform clinical development and improve success rates remains to be demonstrated at scale.

What became clear to me is that we need to be more ambitious about how we define success.

Efficiency matters, of course. But the real measure of AI’s impact should be whether it helps us make better decisions, reduce unnecessary failures and bring better medicines to patients.

The opportunity isn’t simply to do the same things faster. It’s to make better decisions about what we do in the first place.

3. Trust is becoming as important as innovation itself

If there was one word that surfaced repeatedly throughout CONVERGE, it was trust.

Trust in data. Trust in methodologies. Trust in AI-generated evidence. Trust from regulators, clinicians, and patients.

During the Citeline-moderated discussion on synthetic evidence, Professor Shu Yang of NC State University emphasized the importance of rigorous study design, particularly when incorporating external information into clinical trials. The discussion highlighted how innovative approaches can complement traditional randomized trials, provided their assumptions and limitations are properly understood.

Our regulatory panel, moderated by Aloka Chakravarty and featuring Lisa LaVange, Estelle Russek Cohen, and Andrew Thompson, explored what it will take for innovative methodologies and AI-enabled evidence to gain broader regulatory acceptance.

The discussion wasn’t about resisting innovation. It was about defining the intended use, demonstrating credibility, validating methods appropriately, communicating uncertainty and maintaining accountability, and ultimately come with clear use cases.

An important point emerged: not every application of AI carries the same level of risk, and therefore not every application requires the same level of evidence and validation. Understanding the context of use and engaging regulators early can make a real difference.

Eric Tchetgen Tchetgen of the University of Pennsylvania added another important perspective during the closing panel. Even with increasingly sophisticated models and more data, some uncertainty cannot simply be eliminated. There are limits to what we can infer or predict, particularly when trying to understand individual treatment effects.

And that is something we need to acknowledge rather than overlook.

I particularly appreciated that there wasn’t always agreement on how far technologies such as digital twins can take us today. Those differences of opinion are precisely what make scientific discussions valuable.

Trust doesn’t come from the sophistication of a technology. It comes from scientific rigor, transparency, appropriate validation, and a clear understanding of what a method can — and cannot — tell us.

4. Our organizational structures may be holding us back more than our technology

One of the most interesting conversations took place during our Digital Backbone of Clinical Development roundtable, moderated by Kevin Trimm of Cytel and Arnaub Chatterjee of Datavant.

The discussion highlighted an uncomfortable reality: organizations are investing heavily in data, technology, and AI, yet improvements within individual functions don’t necessarily translate into faster or better clinical development overall.

Clinical development, operations, statistics, medical affairs, and market access may each optimize their own activities, while the gaps between them remain largely untouched.

In other words, we may be getting better at individual parts of the process without improving the process as a whole.

We also heard how data fragmentation, organizational silos, and reluctance to challenge established processes can prevent innovative methodologies from being adopted, even when the science is well established.

And there is a human dimension to this challenge that we shouldn’t underestimate.

Justine Rochon also addressed this in her plenary: implementing AI isn’t simply about introducing new tools. It requires leadership, organizational readiness, new capabilities and a willingness to change how people work.

Questions around accountability, ownership of decisions and trust in AI-enabled processes are just as important as the underlying technology.

It may require us to rethink how teams work together, how information flows and who takes responsibility for decisions across the development lifecycle.

Technology can enable transformation. But organizational readiness, leadership, and cross-functional collaboration determine whether that transformation actually happens.

5. We need to move from individual innovation to collective progress

Perhaps the most encouraging theme of the day was the willingness to discuss collaboration beyond organizational and competitive boundaries.

In our closing panel, moderated by Rod MacKenzie, Arnaub Chatterjee, Laura Pizzi of ISPOR, Eric Tchetgen Tchetgen, Abie Ekangaki of Cytel, and Dooti Roy of Boehringer Ingelheim challenged each other on the possibilities and limitations of AI, digital twins and the data needed to support them.

The conversation repeatedly returned to shared challenges: fragmented data, inconsistent terminology, limited access to high-quality evidence, and the need to learn from unsuccessful approaches as much as successful ones.

Several speakers raised the potential of consortium-based initiatives, where industry, academia, technology providers, and regulators could work together on common scientific and methodological challenges.

Eric Tchetgen Tchetgen also raised a particularly important point about learning from failures. We tend to focus on successful applications and positive results, yet there is enormous value in understanding what didn’t work and why.

What I found encouraging was that the conversation moved beyond identifying common problems toward exploring how we might address some of them collectively.

If we are all facing similar questions about evidence credibility, data readiness, and AI validation, how much progress could we make by addressing some of them together?

The future of clinical development will undoubtedly require continued scientific and technological innovation. But it may depend just as much on our willingness to share knowledge, challenge assumptions, and work together on problems that no single organization can solve alone.

Keeping the conversation going

One of the things I appreciated most about CONVERGE was that we didn’t always agree. And frankly, that was part of the intention.

We brought together people with different expertise, experiences, and perspectives, not to confirm what we already believe, but to challenge our thinking.

Did we leave with all the answers? Of course not. But we did leave with a clearer understanding of the challenges we need to address, the opportunities worth pursuing, and, importantly, where collaboration could help us make progress.

What also struck me was how interconnected the discussions became. A conversation about AI inevitably led to questions about data quality, scientific credibility, and accountability. Discussions about innovative trial designs brought us back to evidence requirements, patient involvement, and access. And again and again, we returned to the need to work beyond our individual disciplines and organizational boundaries.

For me, that was perhaps the most important outcome of CONVERGE.

My sincere thanks to all our speakers, moderators, and participants for their openness, expertise and willingness to challenge each other. And, of course, to the team who helped bring CONVERGE to life.

We’re already looking ahead to CONVERGE 2027, and I hope some of the conversations we started in Boston will continue, evolve, and perhaps even lead to new collaborations.

Because ultimately, the question isn’t just what science and technology make possible. It’s what we can achieve together — and what difference that will make for patients.

Share Via