Project Summary

One of PCORI’s goals is to improve the methods that researchers use for patient-centered outcomes research. PCORI funds methods projects like this one to better understand and advance the use of research methods that improve the strength and quality of comparative effectiveness research.

What is the project about?

Knowing which treatments work best for which groups of people can help doctors improve the care they provide. But some research studies don’t have enough participants for researchers to know how well treatments work for people based on their traits, such as age or race.

One way that researchers can get enough data to compare treatments for different groups of people is to combine data from multiple studies. But studies may use different methods and procedures. For example, in experiments, researchers assign people to a treatment. In observational studies, researchers observe how well treatments work without trying to change the treatment patients get. Such differences can make combining the data from different study types difficult.

In this study, the research team is developing new methods for combining data from experiments and observational studies.

How can this project help improve research methods?

Results may help researchers when considering ways to combine data from different types of studies to learn how well treatments work for people with different traits.

What is the research team doing?

The research team is adapting two methods for combining data from experiments and observational studies. The first method is machine learning. In machine learning, computers use data to learn how to perform different tasks with little or no human input. The second method is Bayesian meta-analysis. This method uses prior knowledge and information about the data, such as how accurate or complete it is, to perform the analyses. The team is also developing ways to analyze data when measures of health outcomes vary across data sources.

The research team is testing the methods by comparing two medicines for depression using two sources. The first source is clinical trial data. The second is electronic health records from the Duke University Health System and Johns Hopkins University Health System. The team is looking to see how the effects of these medicines on outcomes such as depression symptoms and hospital visits differ by people’s traits such as age. Finally, the team is developing guidance for other researchers on how to use the methods.

Research methods at a glance

Design Element Description
  • Develop statistical methods to identify heterogeneity of treatment effect using data from combined data sets with both experimental and nonexperimental studies
  • Develop diagnostics and methods that allow researchers to focus on the identification of effect heterogeneity in particular target patient populations
  • Develop guidance for the appropriate use of the methods developed
Approach Machine learning. Bayesian analysis

Project Information

Elizabeth Stuart, PhD
Johns Hopkins Bloomberg School of Public Health
Advanced Computational Approaches for Integrating Data to Assess Effect Heterogeneity

Key Dates

July 2021
November 2025


Award Type
State State The state where the project originates, or where the primary institution or organization is located. View Glossary
Last updated: February 22, 2023