Forensic assays left-censor: results below the limit of detection are
reported only as "< LOD". Naive deletion or zero-filling biases summaries.
This applies a documented simple-substitution rule to the censored entries;
for regression on censored data prefer
morie_horowitz_censored_regression.
Arguments
- values
Numeric vector of measured concentrations; censored entries are
NA(or values belowlod).- lod
Limit of detection (scalar > 0). Entries
NAor< lodare treated as left-censored.- method
"half"(LOD/2, the common default),"sqrt2"(LOD/sqrt(2), for moderate censoring), or"lod"(substitute LOD).
Value
A list with the imputed vector, a logical censored mask,
the fraction_censored among measured values, and n_missing
(NA is not measured, not below the limit; it stays NA).
Examples
morie_tox_left_censor_impute(c(0.4, NA, 0.9, 0.02), lod = 0.05)$imputed
#> [1] 0.400 NA 0.900 0.025