recoverlite - Pre-Data Recovery Tests for Planned Study Designs
A prototype implementation of the pre-data recovery test: a standardized simulation protocol for evaluating whether a planned design-analysis pair can recover its target estimand before data collection. Researchers declare the estimand, data-generating assumptions, data strategy, missing-data process, and analysis strategy; the package simulates a crossed scenario grid (null and target effects, each under declared and pessimistically perturbed nuisance assumptions), applies the planned analysis to each simulated dataset, and reports target-null rejection, power, target bias with its exact decomposition into estimator bias and estimand drift, coverage, Type S and Type M errors, precision, and classified model failure, each with Monte Carlo uncertainty and explicit inclusion rules. A versioned threshold profile converts the diagnosands into a PASS/RISK/FAIL verdict - a decision convention, not a validity classification. Built on the 'DeclareDesign' framework.
Last updated
clinical-trialsdeclaredesignestimandsmetasciencemonte-carlopower-analysispreregistrationresearch-methodssimulationstatisticsstudy-design
2.70 score 1 starsvahtian - Reproducible, Provenance-First Evidence Tooling
Freeze a record set into a content-hashed, provenance-stamped, date-locked corpus; verify reproducibility; and keep a hash-chained, tamper-evident audit trail. The canonical on-disk format and the content hash are byte-identical with the Python package 'vahtian', so frozen corpora and audit ledgers interoperate across the two languages.
Last updated
citationsresearchresearch-integrityresearch-methodszotero
2.48 scorevahtian.epinet - R Interface to the EpiNet Toolkit
An R interface to EpiNet, a transparent toolkit for honestly evaluated outcome models on tabular and graph-shaped data. It wraps the tested Python 'vahtian.epinet' package through 'reticulate', so the algorithms are single-sourced and cannot diverge across languages, and returns calibrated, caveated results: discrimination, calibration, bootstrap intervals, a label-permutation null, and feature importance. A research and education demonstrator, not clinical decision support.
Last updated
calibrationclinical-prediction-modelsdata-scienceepidemiologyfederated-analysismachine-learningnetwork-analysispythonreproducibility
2.30 score 1 stars