Formal causal reasoning is needed at every stage of research — from design, to analysis and interpretation. The CCI focuses on developing and implementing study design and analytic methods applicable to research questions spanning a wide range of fields.
One area where the CCI will play a large role is in precision medicine. In particular, causal inference methods are what is needed to take information from Big Data and turn it in to the evidence required for precision medicine. Whereas machine learning focuses on predicting what will happen if clinical practice stays the same (i.e., predicting whether a new patient will develop the outcome based on historical data), causal inference focuses on predicting what would happen if practice changed (i.e., determining which interventions would be best for which patients when).
MISSION AND SCOPE
Mission 1: Methods Development. The CCI will support the development of novel causal inference methods. Areas of focus include: instrumental variables; matching; mediation; Bayesian nonparametric models; semiparametric theory and methods; propensity scores; structural nested models; estimating optimal dynamic treatment strategies; and sensitivity analysis. The development of these methods also include software, such as R packages, making it easier for researchers to implement the new methods.
Mission 2: Collaborative Research. The promotion of multidisciplinary collaborative research is a major goal of the CCI. Currently, Penn researchers in causal inference play major roles in many fields within the School of Medicine and beyond, including: chronic kidney disease, behavior and addiction, cancer, criminology, critical care, economics, education, genetics, health policy, infectious disease, imaging, marketing, pediatrics, pharmacoepidemiology, political science, and psychology. The CCI is looking to enhance and expand these collaborations.
Mission 3: Education & Training. With the rapid increase in detailed data sources that can be linked (such as electronic medical records, medical claims, imaging, high-throughput biological data, and geocoded-data), there is a corresponding increase in the need for new causal inference theory and methods. In addition, as new statistical methods are developed, there is a need for practitioners (such as epidemiologists and clinical researchers) to be trained in these methods. Thus, the educational mission of the CCI is twofold. First, the CCI will educate statisticians and biostatisticians in causal inference methods, and offer mentorship as they begin independent research in this area. Second, the CCI will educate researchers on the latest causal inference methods: from implementation, to interpretation, to the checking of assumptions and sensitivity analyses.
The (informal) Penn Causal Inference Research Group was formed in 2002. What began as a few researchers interested in discussing the blossoming field of causal inference, grew into a major area of research at Penn.
The research group had weekly meetings from 2002-2016. During that time members of the group developed new statistical methods, trained PhD students and postdocs, hosted 3 conferences, started a journal, and collaborated with clinical investigators in a wide variety of fields. The group included faculty, postdocs, and students from Biostatistics, Statistics, and Epidemiology. In recognition of the productivity and growth of the group, as well of the increasing importance of causal inference methods, the Center for Causal Inference was formed in 2016. In 2018, the CCI became a Penn-Rutgers partnership.