Pharmacometrics¶
PyMixEF provides four connected layers: canonical event data, closed-form PK,
event-aware ODE simulation, and typed pharmacometric model declarations that
compile to the shared pymixef.ir.ModelIR.
Canonical dosing/observation semantics, deterministic same-time order, ADDL/infusion expansion, provenance, numerical integration, and sensitivities.
One- and two-compartment bolus, infusion, and oral profiles; residual-error and censoring likelihood helpers.
Constrained parameters, ETAs, covariates, states, dose mappings, differential equations, observation models, capability validation, and ModelIR compilation.
Random-effect transforms, conditional modes, Laplace population objectives, shrinkage, finite differences, and the exact boundary of experimental SAEM and unavailable FOCEI.
Calculation layers are intentionally distinct
Closed-form and ODE functions simulate from supplied parameters. The DSL declares and validates a model. Conditional-mode/Laplace routines expose estimation primitives. None of those alone is an integrated production population-PK estimator.
For an end-to-end learning sequence, use event semantics, PK and ODE simulation, and DSL to ModelIR.