Back to roadmapOpen Shiny

Causal inference for real-world data

Analytics Lite

A Navigator-informed estimation handoff.

Coming soon: onsite analytics estimation later.

QDM
Define a query to begin
0 of 3 pillars satisfied (Query / Data / Model)
Q · QueryD · DataM · Model

Data Handoff

Point-treatment ATE for this first module.

Drop a CSV here

or choose a local file

Data dictionary

Variable, label, description, role, type, values.

Course material map

2

Current

5

Future

1

Demos

1

Refs

The catalog now separates the working Analytics path from tucked-away teaching demos. The Session 11a CCW Shiny app is the longitudinal Shiny path; Causal Platter stays as an illustration.

Point-Treatment Track

In scope for the current ATE module

IPW lab

Cattaneo birthweight example with propensity scores and SuperLearner refinement.

session
IPWPropensity scoreSuperLearner

materials/sessions/Session 4a/

TMLE simulation

Simulation-based point-treatment TMLE comparison against naive, PSM, and g-computation estimators.

session
TMLESimulationG-computationPSM

materials/sessions/Session 6b/

Longitudinal + Shiny Track

CCW recipe, discretizer, LTMLE, and policies

CCW Recipe Shiny

Interactive clone-censor-weight recipe for immortal time bias: naive analysis, clone, censor, weight, and compare.

app
Target trial emulationCCWIPCWSurvival

materials/sessions/Session 11a/Materials/app/

TTE to LTMLE

Complete lecture bundle linking target trial emulation, immortal time bias, CCW, and LTMLE.

session
TTEImmortal time biasCCWLTMLE

materials/sessions/Session 11a/

CCW vs LTMLE

Research project and Shiny formatter showing CCW as an IPW approximation to the LTMLE target.

project
Per-protocol effectCCWLTMLELongitudinal formatter

materials/project-ccw-vs-ltmle/

Teaching demos

Causal Platter

Teaching demo comparing point-treatment estimators on simulated data; retained as a tucked-away illustration.

app
ATEIPWG-computationCTMLE

materials/apps/causal-platter/

Reference shelf

Targeted Learning books, LTMLE documentation, applied LTMLE tutorial, and modified treatment policy paper.

materials/references/

Co-pilot

Start with the analysis dataset. Nothing leaves the browser in this lite workflow.

  1. 1DataUpload a CSV for the local handoff.
  2. 2Dictionary
  3. 3Map
  4. 4Estimate
  5. 5Export
ConfoundersEstimator Plan

Current scope

Cross-sectional point-treatment ATE.

Next: R Shiny execution, longitudinal discretizer, LTMLE, and clone-censor weighting.

Open Shiny Workbench

https://andystats.shinyapps.io/analytics-lite-r/

Navigator signals

Treatment

Not specified

Outcome

Not specified

Estimand

Not specified

Estimator

Not specified

Confounder candidates

No Step 2 confounders captured yet.