🌐 English
- English
- 简体中文
These self-study notes start with influence functions and build towards doubly robust estimation and targeted maximum likelihood estimation (TMLE). Using the same public SUPPORT/RHC patient data throughout, they connect mathematical derivations, visual explanations and calculations in Julia and R.
| Chapter | What we work out |
|---|---|
| 0 · Getting started | How doubly robust estimation, the EIF and TMLE fit together |
| 1 · The mean | Why the mean influence function is the outcome minus the mean |
| 2 · Perturbing a distribution | How probability changes lead to scores and directional derivatives |
| 3 · Separating the mechanisms | How outcome risk, treatment allocation and baseline characteristics form the target |
| 4 · Deriving the ATE EIF | Where residuals, inverse probability weights and centring come from |
| 5 · Checking every direction | How the gradient represents changes along smooth paths |
| 6 · Finding the efficient gradient | How model restrictions, tangent spaces and projections determine the EIF |
| 7 · Understanding estimation error | How the EIF connects to uncertainty and the efficiency bound |
| 8 · Building AIPW and TMLE | How to construct both estimates from the same patient records |
The worked example follows 5,735 critically ill patients in the SUPPORT/RHC observational dataset. Treatment records whether a patient received right heart catheterisation (RHC) during the first 24 hours; the outcome is all-cause death within 30 days. The same records connect the initial mortality mean to the final adjusted risk comparison.
Patient summary and data sources · References and further reading
Original study: Connors AF Jr, Speroff T, Dawson NV, et al. The effectiveness of right heart catheterization in the initial care of critically ill patients. JAMA. 1996;276(11):889–897. Study record.
Data obtained from hbiostat.org/data courtesy of the Vanderbilt University Department of Biostatistics.