Model capability map

See which engines exist, which failed transfer, and which have not been built.

Model status is reported as a scientific result—not hidden behind a generic “AI” label.

Model 1

Same-context perturbation baseline

Available · limited

Current basis: Trained and evaluated inside the processed HepG2 response dataset.

Result: Small average-error improvement; response direction is not reliable.

Model 2

Tested · unsupported

Current basis: Evaluated on a 55-target external aggregate diagnostic.

Result: The current model did not transfer directionally.

Model 3

Not trained

Current basis: Requires independent DMD and matched-control perturbation outcomes.

Result: No calibrated DMD prediction is emitted.

Model 4

Not available

Current basis: Requires longitudinal patient-linked state, intervention and outcome data.

Result: No patient-specific trajectory or treatment simulation is permitted.