ISPOR 2025 wasn’t just another industry conference – it showcased the health economics and outcomes research (HEOR) community at a critical inflection point. 

As active participants – on stage, in sessions, and in countless hallway conversations – Datavant had a unique view into the most pressing questions and promising breakthroughs in HEOR. This year’s program pushed beyond theory into practice—spotlighting tangible strategies for real-world evidence (RWE) generation, cross-sector collaboration, and patient inclusion in decision-making.

In this recap, we highlight the most compelling themes from the week, share what we learned from conversations across sessions and the exhibit hall, and reflect on how trusted data connectivity is enabling the next generation of evidence.

Whether you made it to ISPOR or are catching up from afar, consider this your curated guide to what mattered most—and what’s coming next.

Global Value Frameworks Are Converging, But Progress is Uneven

Across sessions focused on international health technology assessment (HTA) and regulatory policy, one theme stood out: stakeholders are working toward harmonization, but meaningful convergence remains slow. While global regulators and HTA bodies increasingly accept RWE, the degree of rigor and requirements for data quality, transparency, and relevance vary considerably by region.

As multinational evidence generation becomes more common, we heard growing calls for frameworks that support both global alignment and local relevance. A clear opportunity lies in the development of platform-agnostic, data-native methodologies that allow RWE to be interpreted contextually, not uniformly.

Privacy-Enhancing Technologies (PETs) Are Unlocking More—and Safer—Data Use

As health data becomes more complex and interconnected, the traditional tradeoff between data utility and privacy is being redefined. Privacy-enhancing technologies (PETs) were a recurring theme at ISPOR 2025, not just as compliance tools, but as strategic enablers of secure, scalable evidence generation.

In a panel hosted by Integral, featuring Datavant and AIQ, experts explored how PETs help peel back the barriers that have historically limited data access—while adding intelligent, technical safeguards to reduce the risk of re-identification. Here are a PETs gaining traction across the HEOR ecosystem:

• Synthetic data was highlighted as a way to evaluate datasets prior to purchase or train models without exposing underlying patient information. 
• Data clean rooms allow researchers to query sensitive data without moving it—enabling control, auditability, and protection for data sources. 
• Differential privacy algorithmically adds  “noise” to a dataset in a way that protects individual identities while preserving broader analytic trends. It can be applied on its own or layered into other PETs like clean rooms. While powerful, it was acknowledged to require thoughtful application, especially in rare disease contexts where too much noise can distort insights - resulting in a poor balance between utility and privacy.
• Tokenization allows interoperability between disparate health datasets while removing core elements of PII. Tokenization paired with an expert determination is a method to de-identify data for research use-cases. 

‍The key takeaway? Rather than relying on a single solution, panelists emphasized using PETs in combination to maximize data utility while minimizing risk—laying the groundwork for scalable, privacy-conscious research.

AI in HEOR: Emerging Use Cases, Cautious Optimism

Artificial intelligence, especially “human in the loop” or “AI assisted” workflows, drew significant attention—yet conversations were measured, emphasizing practical applications over hype. In poster presentations and expert panels, AI was framed as a tool to augment, not replace, existing HEOR workflows. 

In a poster presented by Datavant partner Ontada, researchers demonstrated how natural language processing (NLP) can be used to speed access to EHR insights without compromising security or privacy. More specifically, NLP improved data completeness for staging and histology. Unstructured data has long been a focus for richer insights—and AI is unlocking its potential.

Other use cases gaining traction include literature review automation, protocol drafting, and real-time signal detection from large, unstructured datasets. However, speakers consistently stressed the importance of human oversight as new methods of leveraging AI gain traction. The consensus: we’re still in early days, but the right safeguards could unlock enormous value.

Patient-Centered Evidence is Taking Root—But Faces Scaling Challenges

Patient inclusion has moved from an aspirational goal to an operational imperative. Sessions on digital endpoints, advanced patient-reported outcomes (PROs), and the integration of social determinants of health (SDOH) highlighted how stakeholders are building more representative, human-centered datasets.

That said, challenges remain. Standardizing data collection across decentralized sources, integrating patient voice into study design, and ensuring equity in participation will require both technical and governance innovation. Attendees were optimistic, but clear-eyed: inclusion is no longer optional, but execution remains complex.

A Breakthrough Spotlighted by Eli Lilly and Datavant: Cloud-First Insights Generation 

In the Exhibit Hall, Datavant and Eli Lilly unveiled a cloud-first approach for real-world data (RWD) discovery and assessment—aimed at solving long-standing challenges in speed, privacy, and scalability. Powered by AWS Clean Rooms, this approach enables research teams to find, evaluate, and query datasets without moving or exposing underlying data, cutting months off traditional timelines.

Eli Lilly illustrated the impact through a sample cardiometabolism use case, walking through how their team can leverage Datavant Connect powered by AWS Clean Rooms to:

• Search for data sources matching specific filters (e.g., condition area, latency, and population size)
• Run pre-set patient overlap reports to determine how well datasets aligned with their study cohort and assess fill rates for key variables such as MACE events and cost of care
• Execute custom queries within clean rooms, without ever needing to transfer the underlying data

Why This Matters

This solution addresses several longstanding pain points for both data users and data providers:

• Time-to-insight is drastically reduced—from months to days—by removing manual, duplicative steps in discovery and evaluation.
• Data movement is minimized, protecting proprietary datasets and eliminating the risks that come with duplicating sensitive information.
• Visibility and control are improved for all parties, with permission-based access and transparent query history baked into the infrastructure.

For data producers, this also means fewer ad hoc RFPs and more meaningful engagement with qualified opportunities. For life sciences companies, it’s a smarter, faster, and more secure way to answer high-priority clinical and commercial questions. As research teams look to do more with real-world data—faster and more collaboratively—cloud-first infrastructures like this are poised to become the new standard. Learn more here.

Looking Ahead: Innovation Must Be Matched by Infrastructure

ISPOR 2025 affirmed that the tools, frameworks, and ambition to transform HEOR are all within reach. What’s needed now is robust infrastructure—both technical and governance-based—to translate innovation into impact.

In a session hosted by Parexel, experts noted the importance of planning evidence generation throughout the product lifecycle for a cohesive strategy. Frameworks such as integrated evidence plans (IEPs) can enhance overall asset strategy and delivery, particularly in the dynamic landscape of the Inflation Reduction Act (IRA).  Premier also highlighted the role of RWE in all areas of evidence generation – informing site selection, pragmatic trials or those with low interventions, leveraging a mix of secondary data use and primary data capture, and calibrating protocols to solve evidence gaps. 

Looking forward, we expect to see:

• Growth in public-private evidence networks, enabled by PETs
• Increasing regulatory engagement on RWE standards
• Broader adoption of early, advanced preparation of RWE strategies to set clinical research up for long-term success

HEOR is no longer on the periphery of decision-making—it’s at the center. But its continued evolution depends on building the connective tissue that allows stakeholders to collaborate with confidence.

Closing Thoughts

ISPOR 2025 made it clear: HEOR is expanding—not just in scope, but in influence. As expectations for real-world evidence grow, so too must our collective commitment to ethical data use, methodological rigor, and privacy-first innovation.

At Datavant, we remain focused on helping organizations navigate this evolving landscape—by enabling privacy-preserving connectivity, supporting trusted data partnerships, and advancing the infrastructure needed for next-generation evidence.