Derive "Tier 0" variables before any marker function runs.

Computes the following variables only when they are absent from data:

BMI

from weight (kg) and height (m or cm) via weight / height_m^2

glucose / G0

bidirectional alias: whichever is absent is filled from the other

insulin / I0

bidirectional alias (pmol/L <-> muU/mL unchanged – same unit assumed)

eGFR

CKD-EPI 2009 creatinine equation from creatinine, age, sex (and optionally race). Written as eGFR only; downstream functions that need eGFR_cr receive it via col_map.

UACR

urine_albumin / urine_creatinine (both in mg/mmol or the ratio already in mg/g – no unit conversion; caller's responsibility)

LDL_c

Friedewald: TC - HDL_c - TG/2.2 (mmol/L); skipped if TG > 4.5

Usage

.hm_global_precompute(data, col_map = NULL, verbose = FALSE)

Arguments

data

data.frame after col_map materialization.

col_map

Named list of key -> column mappings (used only to resolve the physical column names when materialization has NOT been run yet).

verbose

Logical; emit one-line summary per derived variable.

Value

A list:

data

data.frame with new columns appended.

precomputed

Character vector of variable names that were derived.

Details

col_map keys are respected: if the user has mapped e.g.\ creatinine -> "Cr_serum", the materialized data[["creatinine"]] column (placed there by .hm_build_col_map()) is used transparently.