PBHLTH 243B
Targeted Learning in Practice
This course follows PBHLTH 243A as a two part series. It provides a self-contained introduction to the computational tools for Targeted Learning through it's accompanying software ecosystem, the tlverse. Each class incorporates a concise preliminary lecture, vignette-guided live coding exercises, and discussion. Students will have the opportunity to perform hands-on implementation of novel estimators for answering causal questions with real- world cross-sectional data using the tlverse software ecosystem of R packages.
Topics
- Follow the roadmap of statistical learning to define estimation problems in realistic statistical models, and obtain valid inferences. Apply standard cross-validation schemes using the origami R package, including V-fold, stratified, ad cluster-specific cross-validation. Train a super learner using the sl3 R package by selecting an appropriate loss function, metalearner, and assembling a library of candidate machine learning algorithms. Estimate the effect of a static intervention using the appropriate "Spec", as defined in the tmle3 R package, and apply tmle3's delta method in order to estimate transformations of existing parameters. Use the tmle3mopttx R package to learn the optimal individualized treatment regime, and to estimate effects under such data-adaptive regimes. Approximate causal effects under stochastic treatment regimes with the tmle3shift R package by defining a single shift, grid of counterfactual shifts, or individual-level shifts of the treatment. Estimate direct and indirect effects based on decompositions of the total causal effects of static and stochastic interventions with the tmle3mediate R package. Differentiate stochastic, dynamic, optimal dynamic, and static treatment regimes from each other, and interpret effects under each kind of intervention. Obtain and interpret variable importance metrics from the following tlverse R packages: sl3, tmle3, tmle3mopttx, tmle3shift, and tmle3mediate.
Official catalogue course objectives and student learning outcomes
Sources
- Course catalogue ↗2026-2027