Newswise — Rutgers School of Public Health associate professor, Liangyuan Hu, has received a three-year $1,069,876 Patient-Centered Outcomes Research Institute (PCORI) contract.
Hu will use the funding award to develop robust and flexible causal inference methods and tools, leveraging machine learning, for longitudinal treatments. The developed methods will be used to analyze quasi- or non-experimental data to generate real-world evidence for treatment decisions and intervention policies.
Real-world evidence is essential to answering clinical questions in comparative effectiveness research (CER) and patient-centered outcomes research. In many circumstances, randomized controlled trials are not practical or ethical and their stringent inclusion and exclusion criteria limit generalizability to vulnerable populations such as frail and elderly people, disadvantaged racial groups, and those at risk for severe morbidity and mortality. Drawing causal inference from large-scale data collected from real-world clinical settings is therefore critical to forming important policy related to interventions with patient-centered outcomes.
There is a substantial body of causal inference methods with a time-fixed treatment. In comparison, causal inference methods, particularly flexible ones using machine learning, for time-varying treatment are relatively sparse due to additional complexities associated with time-varying confounding, selection bias and longitudinal data structures. Furthermore, existing approaches in this area no longer meet the growing challenges posed by complex health data structures and treatment patterns.
Hu and her team will propose new methods to fill in these critical methodological gaps in longitudinal causal inference research and to address the important patient-centered outcomes research questions.
“We will develop a suite of robust and flexible longitudinal causal inference tools that will provide a key analysis apparatus to researchers who work with real-world clinical data,” says Hu. “Specifically, we will develop a new, robust marginal structural quantile model to draw simultaneous causal inference about longitudinal treatments across the entire distribution of outcomes and further improve the flexibility of the model by using machine learning. For censored survival outcomes, we will first develop a new, joint marginal structural model, in continuous-time, for the restricted mean survival times and then develop a Bayesian likelihood-based machine learning method that can accommodate time-varying covariates to estimate a set of weights for correcting time-varying confounding or selection bias due to informative censoring. To tackle the ‘no unmeasured longitudinal confounding’ assumption, we will further develop a flexible and interpretable sensitivity analysis framework. Finally, we will develop open-source software within the R computing platform and implement our new methods to solve emerging CER questions in cardiovascular disease research and COVID-19 research.”
PCORI is an independent, nonprofit organization authorized by Congress in 2010 with a mission to fund research that will provide patients, their caregivers, and clinicians with the evidence-based information needed to make better-informed health care decisions.
###
About the Rutgers School of Public Health
The Rutgers School of Public Health is New Jersey’s only accredited school of public health that seeks to improve health and prevent disease in diverse populations in New Jersey and around the world through educating students to become well-qualified and effective public health leaders, researchers, and practitioners; conducting research to advance public health science and policies; and providing service programs that promote population and individual health. Visit us at https://sph.rutgers.edu and follow us on Facebook, Twitter, Instagram, and LinkedIn to learn how we’re “keeping the ‘public’ in public health.”
Source: https://www.newswise.com/articles/rutgers-faculty-receives-pcori-contract-to-develop-robust-longitudinal-causal-inference-methods-with-machine-learning?ta=home