Skip to content

Beta-Alanine Lab: audit pass 2 fixes - #5

Merged
jesse-anderson merged 1 commit into
mainfrom
beta-alanine-audit-pass2
Nov 19, 2026
Merged

jesse-anderson merged 1 commit into
mainfrom
beta-alanine-audit-pass2

Conversation

@jesse-anderson

Copy link
Copy Markdown
Owner

A rest phase at the end of a protocol was counted as dosing: the end-of-dosing
card, back to normal, the maintenance dose and washout cover all started after
the rest. buildSchedule now returns stopDay, the end of the last dose, and the
band, horizon, chart and cards use it.

Rezende 2020's response SD of 6.6 is the spread between people in one lab's
individual data, net of normal variation, so the person factor now adds it to
the spread between papers instead of topping the papers up to it
(CV about 0.46 instead of 0.37). The audit trail says so.

The evidence horizon shades a break between doses longer than any washout
follow-up and flags it, matches a run by its smallest and largest dose so a
drifting ramp is not covered by studies near its top alone, and counts exactly
1.5 times as comparable despite floating point.

A plain serving above the 800 mg tested and under 10 mg/kg is now untested
rather than "possible, usually mild".

Six claims from Rezende 2020 and Stellingwerff 2012 back figures the page
printed without one. The audit and anchors read them, a spec test holds every
anchor's measured figures to its claims, and a new anchor puts Rezende's
0.5 mmol/kg dm a week after week 4 against the model (model above, the slowdown
gap). Changing the washout shown keeps a study preset, and the person factor's
CV band runs once per page.

…dit figures to claims

A rest phase at the end of a protocol was counted as dosing: the end-of-dosing
card, back to normal, the maintenance dose and washout cover all started after
the rest. buildSchedule now returns stopDay, the end of the last dose, and the
band, horizon, chart and cards use it.

Rezende 2020's response SD of 6.6 is the spread between people in one lab's
individual data, net of normal variation, so the person factor now adds it to
the spread between papers instead of topping the papers up to it
(CV about 0.46 instead of 0.37). The audit trail says so.

The evidence horizon shades a break between doses longer than any washout
follow-up and flags it, matches a run by its smallest and largest dose so a
drifting ramp is not covered by studies near its top alone, and counts exactly
1.5 times as comparable despite floating point.

A plain serving above the 800 mg tested and under 10 mg/kg is now untested
rather than "possible, usually mild".

Six claims from Rezende 2020 and Stellingwerff 2012 back figures the page
printed without one. The audit and anchors read them, a spec test holds every
anchor's measured figures to its claims, and a new anchor puts Rezende's
0.5 mmol/kg dm a week after week 4 against the model (model above, the slowdown
gap). Changing the washout shown keeps a study preset, and the person factor's
CV band runs once per page.
@jesse-anderson
jesse-anderson merged commit db24115 into main Nov 19, 2026
2 checks passed

This branch was successfully deployed

1 active deployment
github-pages — db24115d Deployed Nov 19, 2026 by jesse-anderson via deploy #177
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant