“Our goal is to build a powerful, more accurate way of illuminating the racial disparities in procedures like arthroplasty, including knee or hip replacements,” said Yan Ma, PhD, MA, MS.
Researchers Aim to Create Machine-Learning Method to Better Understand Racial Disparities in Health Care
WASHINGTON, D.C. (Oct. 21, 2019) – Milken Institute School of Public Health (Milken Institute SPH) at the George Washington University today announced a $1 million grant from the National Institute on Minority Health and Health Disparities to use machine-learning techniques to create an innovative way to better understand racial disparities associated with common surgical and medical procedures.
Yan Ma, PhD, MA, MS, an associate professor and Vice Chair of the Department of Biostatistics and Bioinformatics at Milken Institute SPH, will serve as the principal investigator of the four-year project.
“Our goal is to build a powerful, more accurate way of illuminating the racial disparities in procedures like arthroplasty, including knee or hip replacements,” Ma said. “Ultimately, we hope our machine-learning method will improve the quality of research on all kinds of health care disparities.”
Ma and his colleagues knew that the large national databases currently used to study racial disparities in health care typically were missing key data, such as information on the patient’s ethnic group or race. As demonstrated in their prior work, missing race has implications for the accuracy of other patient characteristics such as older age, length of stay, emergency admissions, anesthesia type and payer. To address this problem, the team plans to use novel statistical methods based on machine-learning techniques that can impute the missing data.
Once the team has created more precise data sources, they will use them to assess racial disparities such as having a higher risk of dying or complications after undergoing total joint arthroplasty, one of the most common surgical procedures in the United States.
As the American population ages, many will need arthroplasty, often because they suffer from osteoarthritis of the knee or hip. Total knee arthroplasty (TKA) and total hip arthroplasty (THA) are considered safe, long-term, cost-effective treatments for osteoarthritis. In fact, more than 7 million Americans were living with a TKA or THA in 2010.
In the coming decades, the demand for arthroplasty is expected to rise. As TKA and THA can potentially improve the health of many Americans, there is a great need to understand, track and resolve racial and ethnic disparities that often go along with these procedures, Ma said.
Past studies have suggested that minority populations have lower rates of total knee replacement utilization, but higher rates of adverse health outcomes associated with the procedure. At the same time, it is hard to get an accurate picture of the problem if some of the data underlying the research is missing, Ma said.
“In order to better elucidate factors that underlie disparities in access to care, a major social determinant of health, data sources used to measure disparities must be reliable and consistent,” Ma said.
Although this study will focus on using the new method to study racial disparities in arthroplasty, in the end the team hopes to have a better tool that researchers can use to study other types of racial disparities.
For example, past research has suggested blacks and Latinos are more likely to develop or die from certain health problems like cancer. Yet the missing data on race in large national databases affects research on these conditions as well.
“Using machine learning to impute missing data on race we hope to create a better method of studying all types of racial disparities in health care,” Ma said. “Armed with better information about the cause and prevalence of such disparities, researchers can begin the search for solutions to this problem."