AIPW (Augmented Inverse Probability Weighting)

Use the synthetic data setup first; it defines data and the columns used below. These are advanced direct R examples.

Combine outcome and propensity models, with optional cross-fitting. Double robustness still requires the other identification and regularity assumptions. For binary outcomes, the current estimator reports risk differences, not ORs.

aipw_result <- causal_aipw(data, "Immunotherapy", "Response",
                           confounders = c("Age", "Sex", "Stage", "ECOG"),
                           outcome_type = "binary",
                           crossfit = TRUE,  # Cross-fitting to reduce overfitting
                           K = 5)

Read the result

For non-survival outcomes, current outputs are RiskDiff or MeanDiff. Cross-fitting does not remove the identification assumptions. Continuous-exposure policies and survival contrasts need explicit eligible settings; the simple binary-exposure example does not establish their validation.

See effect scales and reporting, diagnostics and the current support reference.