v2 · R
A published finding, rebuilt from public WHO data instead of trusted on the page, with the parts that would not reproduce cleanly left visible rather than smoothed over.
nix run github:gauranii/GapYears
Read moreGitHub
v0.1 · R
Two measures of biological age, trained toward the same target, disagree about race in the same people. A replication in public data, and an attempt to take the disagreement apart.
nix run github:gauranii/BloodVsMethylation
Read moreGitHub
v1 · Python
A famous field experiment rebuilt in Python from the authors' raw data, checked against their own Stata log to the last printed digit, then turned into the yes-or-no a campaign actually needs.
Read more
v1 · Python · R · Lean
The primary result of a public Alzheimer's trial submission, rebuilt from raw data in Python and in R, proved value by value in Lean, then refitted the way a trial would handle dropouts today.
Read more
v1 · Python · R · Lean
The 1954 Salk polio trial ran a randomized design and a cheaper observed one side by side. A rebuild of the 1955 report's arithmetic in Python and R, and the 9-point gap between the designs taken apart.
Read more
v1 · Python · R · Lean
The trial behind the 1990 approval of levamisole plus 5-FU for colon cancer, rebuilt from raw data in Python and in R, the verdict proved in exact arithmetic, and the mismatches with the paper left in view.
Read more