Open-source methods and software
The methods layer of Vahtian, published as open source. A method you can read is a method you can check, so each package ships its assumptions, its tests, and an archived release you can cite by DOI. Install them, run them on your own data, and read exactly what they do and do not establish.
RecoverLite R · Python
Pre-data recovery tests for planned study designs: simulate whether a planned design–analysis pair can recover its target estimand, with PASS / RISK / FAIL verdicts under versioned threshold profiles. The Python package is protocol-identical to the R package; results agree within Monte Carlo error.
R: 10.5281/zenodo.21195920 · Python: 10.5281/zenodo.21195922
AssessLite R · Python
Structural assumption assessment for causal analysis. Every causal result borrows strength across units, places, or times, licensed by invariance assumptions that usually stay implicit. AssessLite makes them explicit, attacks them, and records what survived.
pip install assesslite
AuditLite Python · R
A reproducibility audit trail for research projects: records analysis runs, checks for stale outputs, and flags claims downstream of changed sources via multi-hop change propagation — the case a simple timestamp check misses.
EpiNet Python
A transparent toolkit for graph-shaped research data: honest model evaluation (calibration, permutation nulls, model cards), closed-form contestability, and federated governed aggregates. A research demonstrator, not clinical decision support.
vahtian Python · core library
The synthesis core: citation-true research synthesis with Zotero and PubMed intake, provenance tracking, and blinded human–AI dual-rating records. The engine under the CiteVahti workflow.
pip install vahtian
QualiVahti scripts Python · R · Apache-2.0
The open analysis engine of QualiVahti Local: scripts for local-first qualitative analysis — transcription, coding support, codebook exports, and evidence maps that stay on your machine.