HTA, Digital Twins, And Building Clinical Tech That Lasts

This is part three of a three-part series based on a conversation with Rob Abbott, CEO of ISPOR, about the 2026-2027 Top 10 HEOR Trends Report. Part one covered HEOR fundamentals and AI governance. Part two examined real-world evidence, wearables, and data infrastructure.

The EU HTA Regulation went live in January 2025. Joint Clinical Assessments began that same month for oncology and advanced therapy medicinal products. Joint Scientific Consultations, which allow sponsors to pre-consult multiple HTA bodies during trial design, are now available. For anyone designing clinical technology platforms today, that sequence of events isn’t background noise. It’s a direct signal about when HTA requirements need to enter the protocol development process.

The answer, according to ISPOR CEO Rob Abbott, is: earlier than you think.

HTA Belongs Upstream – And So Does HEOR

“I think they should be involved as far upstream as they can,” Abbott told me when I asked when HTA requirements should start influencing data management and statistical analysis plans. “There’s an interesting dance between companies that hope to gain market access for a particular technology and the HTA bodies that are rendering a decision.”

That dance is getting more complicated. HTA bodies, and Abbott confirmed this, are sometimes more stringent than regulators, with higher expectations for the evidence package and specific views about what data they need to see. Many new therapies and technologies struggle with market access, not because they failed regulatory review, but because the evidence generated during development didn’t address what HTA bodies required. The clinical data were there. The HEOR data wasn’t.

The implication for ClinOps and platform design is direct: HEOR professionals need to be engaged upstream and need to understand they’re not working in isolation. “They are part of an entire drug development pipeline,” Abbott said, “and really need to understand that in ways that perhaps historically they haven’t, they’ve focused on what they were tasked with as opposed to how that fits into and facilitates market access down the road.”

What does that look like technically? It means statistical analysis plans need to account for HEOR endpoints – quality of life measures, resource utilization, cost data – from the start. It means data management systems need to be capable of capturing the variables HTA bodies will ask for. And it means that AI in the HTA context – automating systematic reviews, literature screening, data extraction – could “dramatically reduce the time to prepare an HTA submission,” as Abbott put it, but only if the underlying evidence has been designed with HTA requirements in mind.

Digital Twins: Promising, Early, And Moving Fast

The evidence base for digital twins in regulatory submissions is nascent. Abbott was direct about that, but he also was direct about the trajectory.

“This is definitely an earlier stage,” he said. “But given how fast we have seen some of the developments elsewhere, most notably with AI, I think that digital twins and the use of digital twins to inform healthcare decision making hold considerable promise. I would expect to see significant progress over the next three to five years.”

The concept is compelling in a way that maps well to clinical trial technology: a digital twin is a virtual model of an actual patient that mirrors biology, disease progression, and response to treatment, enabling experimentation without risk to the real patient. It strengthens synthetic control arms. It extends what’s possible in rare disease trials where small patient populations make traditional RCT design impractical.

The milestones Abbott identified as necessary before broader acceptance are instructive: more successful pilot projects, a combination of clinical data with genomics, biomarkers, and real-world data to validate the approach, and “just more awareness, understanding, and acceptance of the concept of a digital twin as a particularly useful surrogate for the real patient.” That last one isn’t technical; it’s cultural, and in clinical research, cultural shifts often take longer than technical ones.

For clinical technology builders, digital twins aren’t a procurement decision today. But they’re a design consideration. Platforms being architected now will either be positioned to incorporate digital twin data flows when the evidence base matures, or they’ll need significant rework to do so.

Software As A Medical Device: The Continuous Update Problem

The FDA has now approved 10 mental health apps as Software as a Medical Device, including Rejoyn for depression, EndeavorRx for ADHD, and NightWare for PTSD and sleep. That’s notable in itself. What’s more notable for clinical technology is the structural problem SaMD creates for traditional evidence frameworks: these tools are continuously updated after approval, which means the version evaluated in a pre-approval study isn’t necessarily the version patients are using six months later.

“We’re on the threshold of something really exciting,” Abbott said. “We’re poised to transition from what I might characterize as a static point-in-time assessment to continuous living life cycle approaches to assessment.”

In practice, that means continuous evidence generation, technology reassessments triggered by new clinical and real-world data, and managed access models where time-limited reimbursement lets high-promise technologies enter the market while further evidence is collected. Abbott expects HTA bodies to engage earlier in the R&D cycle, as NICE has begun doing with early value assessments, guiding sponsors on what evidence they’ll need to see in an initial assessment or reassessment.

For clinical technology platforms, this is an architectural challenge. A platform designed around discrete studies with defined endpoints and fixed versions of the tools being evaluated is different from a platform designed to support continuous evidence generation and iterative reassessment. The former is built for a world that’s already changing. The latter is built for the one we’re moving into.

The One Change Abbott Would Mandate

I asked Abbott a closing question: if he could mandate one change in how clinical technology platforms are built and procured, what would it be?

The answer, without hesitation, was HEOR.

“I would really want to see much greater awareness, understanding, and use of health economics and outcomes research in the consideration of these technologies,” he said. “Because I think that is going to ultimately enable us to get to the outcomes that we want.”

It’s a through-line across all three articles in this series. AI adoption without HEOR-informed governance creates the hallucination problem, the data quality problem, the Wild West scenario Abbott warned against. RWE platforms built without HEOR endpoints built in produce evidence packages that HTA bodies can’t use. SaMD evaluated under static frameworks can’t keep pace with continuously updated products. In each case, HEOR is the missing structural element, not as a compliance requirement, but as an orientation toward value that should inform every design decision.

“We are working in the service of better health outcomes,” Abbott said. “We want the best that medical science has to offer at a reasonable cost. So how do we thread that needle? That’s really where some of these emergent technologies come in.”

Wearables, remote monitoring, digital diagnostics, AI, all of it integrated into HTA, shifting evidence from periodic point-in-time visits to continuous real-time patient outcome monitoring. That’s the infrastructure the industry is trying to build. The question is whether the clinical technology platforms being designed and procured right now will be capable of supporting it.

“We are at or remarkably close to an inflection point,” Abbott said. “How we navigate these next few years is going to be really pivotal in terms of the kind of healthcare ecosystem that we’re creating.”

The HEOR community, working through ISPOR, is trying to define what that navigation looks like. The clinical technology community – the people building and buying the platforms that will carry it out – needs to be in that conversation.