Bayesian evidence synthesis, from paper to posterior.
Two pieces, one seam: klin turns published evidence into Bayesian analyses; klindb curates the treatment effects those analyses run on.
klin
Point it at a published paper, a CSV, or a screenshot of a forest plot. klin extracts the data and re-runs the analysis as a Bayesian random-effects model — shrinkage, credible intervals, and the prediction interval the paper never reported. It also produces a defensible estimate from a single trial, using empirical-Bayes priors derived from 1,635 Cochrane meta-analyses.
klindb
A curated relational database of treatment effects from Phase 3 randomized trials, starting with oncology and cardiology. Every effect links back to its source, with versioned snapshots and empirical heterogeneity priors attached.
The single-trial method follows van Zwet, Więcek & Gelman (2025), “Meta-analysis with a single study,” Statistical Methods in Medical Research 34(12) — combining one trial’s estimate with priors estimated from the Cochrane Database of Systematic Reviews to correct the winner’s-curse inflation of a lone significant result.
Request beta access
klin is invite-only during the beta. Leave your details and we'll send you an invite code.