Score the DA-cVAE single-sample disease discriminator

Scores each row of X independently with the frozen fit_dacvae model, in PURE R (no python, no platform/cohort input). Each query is mapped to its standardised Stage-1 structural-ILR balances, embedded by the exported encoder to the deterministic latent \(z = \mu\), and scored by the frozen disease logit \(w^\top\mu + b_0\). Larger = more case-like. Specimens with empty positive support over the frozen universe, fewer than hp$min_features universe features present, or a degenerate (no-balance) model return the neutral score 0. The score of a row depends only on that row and the frozen model and is exactly invariant to per-specimen positive scaling (singlesample_assert_row_equivariant).

Usage

score_dacvae(model, X, meta = NULL)

Arguments

model

A dacvae_model from fit_dacvae.

X

Numeric matrix (samples x features) of non-negative abundances; named cols.

meta

Optional per-sample metadata; accepted for interface uniformity, ignored (the platform adversary is FIT-only).

Value

Finite numeric vector of length nrow(X); higher = more case-like.

See also

fit_dacvae, score_dacvae_novelty