Problem and Need for the Study

Individualized medicine requires risk prediction tools to guide the use of interventions and preventative measures. These tools are not equitable unless they are tailored to different racial and ethnic groups. Risk prediction tools based on mainly white populations may lead to misdiagnosis and poor health outcomes for nonwhite patients.

Developing risk models for minority groups based on local health system data is difficult as they are often underrepresented, leading to a small sample size. The All of Us program offers whole genome sequencing, Electronic Health Record, and patient reported outcomes data from a diverse cohort of patients who are underrepresented in biomedical research. 

Innovation and Impact

Our goal is to use All of Us data to develop new methods for risk modeling that are tailored to minority groups. We are going to start by creating a risk prediction model based on All of Us data for rheumatoid arthritis and cardiotoxicity. The representation of nonwhite groups in this data will allow us to create a model that is more accurate and fair across racial and ethnic groups. The risk models will be evaluated and modified to work with health system data instead of All of Us data.

This project will lead to new methods for researchers to work with All of Us data and demonstrate the potential impact this program can have on healthcare.

Key Personnel and Performance Sites

University of Minnesota

  • Multiple Principal Investigators: Rui Zhang, Jue Hou, Jinhua Wang
  • Research Scientists: Yu Hou, Yinzhao Wang

Harvard Medical School

  • Co-Investigator: Tianxi Cai

This National Institute on Minority Health and Health Disparities grant (R21MD019134) is a two-year, $435,723 award.
Project dates: 25-September-2023 to 31-May-2025