Artificial intelligence is moving quickly into healthcare research, but one question still comes up: can you trust it? If an AI platform is going to support study design, cohort definition, analysis and reporting, RWE teams need confidence that the outputs are accurate, transparent and reproducible.
That is exactly what the new validation study, Validating HealthVerity eXOs: An agentic AI platform for real-world evidence, set out to measure. Using the ISPOR ELEVATE-GenAI framework, the study evaluated HealthVerity eXOs, an agentic AI buit for RWE generation, across 250 validation runs to assess validity, transparency and reliability in AI-generated real-world evidence.
What is HealthVerity eXOs and how does it support AI-generated real-world evidence?
HealthVerity eXOs is an agentic AI platform built to help researchers turn plain-language questions into real-world evidence analyses. It is designed to support common RWE and HEOR use cases, including cohort feasibility, prevalence and incidence, adherence and time-to-event analyses. The platform also gives users visibility into the logic, study design and code behind each analysis, which is critical for teams that need more than a quick answer.
For researchers searching for AI tools for healthcare research, AI for HEOR or agentic AI for epidemiology, this matters. Speed alone is not enough. Trust, transparency and methodological rigor matter just as much.
How was HealthVerity eXOs validated for healthcare research and HEOR?
The validation study used the ISPOR ELEVATE-GenAI framework to assess HealthVerity eXOs across key dimensions such as accuracy, comprehensiveness, factuality and reproducibility. Researchers ran a test set of 50 different HEOR questions five times each, for a total of 250 runs. These questions covered a wide range of research areas, including prevalence and incidence, cohort construction, treatment utilization, comparative effectiveness, adherence, survival and care patterns.
This is important because teams evaluating AI for real-world data analysis want evidence that the platform has been tested across meaningful, research-relevant scenarios, not just narrow examples.
Is HealthVerity eXOs accurate, transparent and reproducible?
According to the study, yes. HealthVerity eXOs showed more than 90% agreement with published literature, zero observed hallucinations, 99.2% code adherence and a median reproducibility score of 0.758 across repeated analyses.
These findings speak directly to some of the biggest concerns around generative AI in life sciences and healthcare analytics. Can the platform produce answers that align with established evidence? Can it follow methodological instructions? Can researchers trust the results to stay consistent across repeated runs? This study suggests HealthVerity eXOs performs strongly in each of these areas.
Why validation matters for AI in real-world evidence
As AI adoption grows in healthcare, many teams are asking similar questions: What is the best AI for real-world evidence? Can generative AI support HEOR research? Is there a transparent AI platform for epidemiology and observational studies?
Validation matters because it helps answer those questions with evidence, not hype. In this study, HealthVerity eXOs also showed strong alignment with the EQUATOR STROBE guidelines used in observational research reporting. In 92% of runs, the platform reported at least 60% of relevant STROBE items, and in 74% of runs, it reported at least 70%.
That gives RWE, HEOR and market access teams stronger reason to trust that the platform can support structured, research-ready outputs.
Human-in-the-loop AI for healthcare research still matters
One of the most important takeaways from the study is that trustworthy AI in healthcare should not operate like a black box. HealthVerity eXOs is built as a human-in-the-loop platform, allowing researchers to review definitions, code sets, analysis plans and outputs before moving forward. The paper also notes that valid analysis depends on fit-for-purpose data, not just strong AI performance.
For life sciences teams, this is a practical reminder that AI works best when paired with expert oversight and high-quality underlying data.
Read the full HealthVerity eXOs validation study
HealthVerity eXOs makes a strong case for transparent, reproducible and credible AI-powered real-world evidence generation. For teams exploring agentic AI for healthcare research, this study offers a valuable look at how validation can build trust in AI-supported analysis.
Read the full validation study to explore the methodology, framework and findings behind HealthVerity eXOs.
