Postdoctoral Fellow

The postdoctoral fellow will work on an externally funded methods development project focused on causal inference for long-term pharmacotherapy. The central methodological contribution is a framework for "patient-choice (PC) protocols": a class of treatment strategies that allow patients to flexibly balance quality of life and clinical outcomes during sustained pharmacotherapy. The applied context is adjuvant endocrine therapy for patients with breast cancer. The fellow will contribute to theoretical development, estimation, and real-world application of these methods across large healthcare claims databases (Merative MarketScan, SEER-Medicare) and data from an international phase-III clinical trial (PALLAS).

Under the general guidance of the Principal Investigator, the postdoctoral fellow will:
• Develop and formalize causal estimands using counterfactual theory, causal directed acyclic graphs, and single world intervention graphs.
• Derive efficient influence functions and develop targeted minimum loss-based estimation (TMLE) algorithms for a general class of PC protocols, including extensions to dynamic regime marginal structural models.
• Implement doubly robust, semiparametrically efficient estimators that incorporate flexible machine learning algorithms for confounding control in high-dimensional longitudinal data.
• Develop and apply sensitivity analysis methods.
• Analyze large-scale observational claims data and contribute to open-source software development in R.
• Prepare manuscripts and present at national and international conferences.

All duties and responsibilities are carried out in compliance with institutional policies, ethical research standards, and applicable federal and state regulations.

LEARNING OBJECTIVES
• The fellow will develop expertise in a novel class of causal estimands for longitudinal treatment strategies with non-adherence, building from foundational identification theory through to efficient nonparametric estimation and open-source implementation.
• The fellow will gain hands-on experience applying causal inference methods in large-scale healthcare claims data.
• The postdoctoral fellow will participate in a weekly causal inference conference with opportunities to collaborate on methods development and applied projects in causal inference research.
• The fellow will develop an independent publication record through authorship on methods and applied manuscripts, and will have opportunities to present work and lead workshops at major statistical and epidemiologic conferences.

ELIGIBILITY REQUIREMENTS
• Ph.D. in Biostatistics, Statistics, or Epidemiology with a strong focus on causal inference methods
• Strong programming skills in R and SAS required, with experience using large administrative claims datasets