Public Health Researcher Awarded $1 Million for Improved Patient-Centered Outcomes Research
Daniel J. Feaster, Ph.D., associate professor of public health sciences, has been awarded $1 million for a three-year project to improve methods for patient-centered outcomes research.
Funded by the Patient-Centered Outcomes Research Institute, Feaster’s project, “Methods for Heterogeneity of Treatment Effects: Random Forest Counterfactual Machines,” was one of 46 new awards selected from 490 submissions.
Feaster captured the importance of the research in one statement.
“Most patients want to know if a treatment will work for them, not whether it works for the average patient,” said Feaster.
Both behavioral prevention and medical research have frequently shown prominent subgroup interactions in the effects of treatment. However, subgroup analysis of clinical trial data is controversial, largely due to increased likelihood of error and lack of replicability. Machine learning techniques, such as random forests, a data mining tool, provide a structured approach to exploring a large number of predictors and identifying replicable sets of predictive factors. In recent innovations, these techniques have been used specifically to uncover subgroups with different treatment responses.
Feaster and his colleagues, including Hemant Ishwaran, Ph.D., M.Sc., professor of public health sciences and Director of Statistical Methodology in the Division of Biostatistics, J. Sunil Rao, Ph.D., professor of public health sciences and Director of the Division of Biostatistics, and graduate students from the new Ph.D. program in Biostatistics, will improve this approach by developing a comprehensive methodology for reliably estimating patient subgroup treatment effects. The methods use optimal person-specific counterfactual random forest machines and are expected to work for heterogeneous and potentially confounded data.
“This research is extremely important because it will provide the tools needed to determine the impact of preventive and treatment interventions on specific subgroups,” said José Szapocznik, Ph.D., professor and Chair of Public Health Sciences and Director of the Miami Clinical and Translational Science Institute. “There are many subgroups that will benefit from this research. One area in which this research will have a tremendous impact is on gender and health disparities research, where we will be able to determine if a particular intervention works with specific gender and/or racial and ethnic subgroups without having to power the study for each subgroup, which would add considerable cost to the trial.”
The funding also will be used to develop free software to implement the new methods in an array of comparative effectiveness research applications and to work with public health and infectious disease clinicians, as well as at-risk patient groups, to incorporate data collection and predictive modeling as part of clinic procedures.
The methods developed in the study, Feaster says, will apply to nearly all comparative effectiveness research by providing a method and software to identify groups for which specific interventions work best.
“These methods have promise to become the backbone of patient information-focused feedback systems, providing patient-specific estimates of expected level of success with alternate treatments,” Feaster said.
The Patient-Centered Outcomes Research Institute is a non-profit organization authorized by Congress in 2010 to fund comparative clinical effectiveness research that will provide patients, their caregivers and clinicians with evidence-based information to make better informed health and healthcare decisions.