Fits an MBMM analysis in which the productive-infection class is represented by a mechanistic trajectory bank and the non-establishment class is represented by an analytic observation model below the assay LOD.
Usage
mbmm_productive_infection(
obs_list,
productive_bank,
sigma,
formula,
data,
lod_fraction = 0.5,
productive_class = "productive",
nonestablishment_class = "nonestablishment",
ridge = 1e-04,
penalty = "none",
n_perm = 0,
seed = NULL,
id_col = "id",
donor_col = "id",
bundle_col = "bundle_id",
log_scale = FALSE,
donor_weighting = c("equal", "particle"),
loo_productive = FALSE,
verbose = TRUE,
maxit = 4000
)Arguments
- obs_list
Named list of subject longitudinal observations.
- productive_bank
Predicted productive-infection reference bank.
- sigma
Measurement-error SD on the Gaussian likelihood scale.
- formula
One-sided gate-model formula.
- data
Subject-level gate covariate data.frame.
- lod_fraction
Expected non-establishment mean as a fraction of LOD.
- productive_class
Label for the productive-infection class.
- nonestablishment_class
Label for the non-establishment class.
- ridge
Nonnegative ridge stabilization parameter.
- penalty
Either
"ridge"or"none".- n_perm
Number of optional gate permutations.
- seed
Optional permutation-test seed.
- id_col
Subject-ID column in
data.- donor_col
Donor-ID column in
productive_bank.- bundle_col
Bundle-ID column in
productive_bank.- log_scale
Whether observations, predictions, and LOD values are already on the log10 likelihood scale.
- donor_weighting
Either
"equal"or"particle".- loo_productive
Whether to exclude the scored subject from the productive bank when that subject is also a bank donor.
- verbose
Print scoring progress.
- maxit
Maximum gate optimizer iterations.
Details
This structure is appropriate for infection-establishment applications in which productive infection produces structured longitudinal kinetics, whereas non-establishment or transient/abortive detection is represented by a low-level analytic likelihood rather than a second ODE trajectory bank.
The expected measurement under non-establishment is defined as a specified
fraction of the observation-specific LOD. A value of 0.5
corresponds to an expected natural-scale measurement of LOD divided by two.
For binary models, gate coefficients describe the log-odds of the
productive_class relative to the
nonestablishment_class.