Using Real-World Data to Answer Real Questions About Children’s Health

For many pediatric conditions, researchers still lack the evidence needed to answer critical questions about treatment effectiveness. Traditional clinical studies can be difficult to conduct in children, particularly for rare diseases and small patient populations. As a result, researchers are increasingly exploring new ways to generate reliable evidence from the data already being collected through routine care.

Children remain one of the most underrepresented populations in clinical research. Traditional randomized controlled trials are considered the gold standard for generating evidence, but in pediatric populations they can be difficult to conduct, expensive to sustain and, in some cases, ethically or practically challenging. For rare diseases, emerging therapies and highly diverse patient populations, gathering enough participants to answer critical research questions can be especially difficult.

At the same time, health systems are generating an unprecedented amount of real-world clinical data through routine care. These data hold valuable insights, but turning them into scientifically credible evidence requires sophisticated methods, significant computing power and specialized expertise.

That challenge is at the center of a new Battelle initiative recently recognized by Amazon Web Services (AWS) through its 2026 AWS Imagine Grant Children’s Health Innovation Award. The award will support Battelle’s efforts to develop new approaches for generating evidence from real-world pediatric health data, helping researchers answer important treatment questions when traditional studies are not feasible.

A Different Way to Generate Evidence

The long-term goal of pediatric research is straightforward: ensure children receive care informed by evidence that reflects their unique biology, experiences and clinical circumstances.

Real-world healthcare data can play an important role in reaching that goal. When analyzed using scientifically rigorous methods, observational data can help researchers better understand how treatments perform across pediatric populations, identify differences among patient groups and estimate the uncertainty that naturally exists in clinical research.

Battelle is working to make these types of analyses more accessible through a cloud-based capability that combines advances in artificial intelligence, causal inference and reproducible analytics.

Rather than focusing on AI alone, the effort brings together several complementary scientific disciplines. The goal is not simply to apply AI to healthcare data, but to help researchers generate scientifically rigorous evidence about treatment effectiveness. By integrating AI with causal inference, uncertainty quantification, fairness assessments and transparent, reproducible workflows, Battelle aims to help researchers better understand how treatments perform across pediatric populations and how much confidence can be placed in the results.

Researchers are exploring ways to embed these capabilities into repeatable analytical workflows that can help investigators generate, evaluate and communicate evidence from observational healthcare data more efficiently and consistently.

Building on Battelle’s Research Foundation

The project builds on years of Battelle research in graph-based machine learning, causal AI and treatment-effect estimation. Researchers have already established foundational scientific frameworks, methodological requirements and governance concepts for trustworthy AI-enabled pediatric evidence generation. Battelle teams have also evaluated causal methods using engineered datasets with known outcomes and controlled levels of confounding, helping validate how these approaches perform before applying them to real-world pediatric research questions.

The current effort is focused on developing a proof of concept that can help pediatric researchers move from a research question to reproducible analytical results through a more streamlined process. If successful, the approach could help generate evidence for important research questions when conventional studies are difficult to conduct.

Collaboration for Greater Impact

Improving pediatric evidence generation requires expertise from across the healthcare and research ecosystem.

Battelle plans to build on existing relationships with children's health organizations, clinicians, health systems, data holders, methodologists and technology partners to evaluate the capability against meaningful real-world use cases. These collaborations will help ensure that the methods remain grounded in clinical needs, that results are interpretable to researchers and clinicians, and that the capability continues to address important pediatric research questions over time.

The project's architecture is also being designed with data stewardship in mind. Rather than requiring large-scale movement of sensitive patient information, the approach is intended to support secure analysis within or near healthcare data environments, allowing organizations to retain control of pediatric data while participating in collaborative research efforts. Summaries, estimates and models can be exchanged without sharing identifiable patient-level data, helping balance local data stewardship with collaborative research and analysis.

Recognition That Advances the Work

Amazon Web Services named Battelle one of 11 recipients of the 2026 AWS Imagine Grant Children's Health Innovation Award. AWS described the project as a pilot research solution that uses artificial intelligence and causal inference to estimate treatment effects from real-world healthcare data.

The award provides funding, computing credits and technical guidance to help Battelle translate its causal AI research into an operational pediatric evidence-generation capability. Over the next year, the team will focus on advancing the project's architecture, data preparation, causal AI deployment, generative AI-assisted workflow support and pilot testing with pediatric research use cases.

While the recognition is an important milestone, the larger opportunity lies in demonstrating a practical and scientifically rigorous way to generate pediatric evidence when conventional studies are difficult or infeasible. If successful, the effort will create reusable architecture, workflows and methodological assets that can support additional research questions and help strengthen the evidence available for children's healthcare decisions.

Read the AWS announcement and learn about the 2026 award recipients: 11 organizations, one mission: AWS announces the 2026 Children’s Health Innovation Award recipients | AWS Public Sector Blog