Simple formulas, room for custom models
This deck shows how BayesianRegressionModels.jl combines concise regression formulas with custom model code. A population-PK example joins subject effects, a concentration calculation, and repeated measurements in one declaration. The larger examples show how this approach extends to more involved scientific models, with their current verification limits stated.
The deck also includes the corrected HSGP centering study and a possible future Julia-to-XLA path for GPU execution through Reactant. That path is a direction to investigate, not an existing BRM integration or a performance claim.
The main talk ends explicitly before a six-slide appendix covering execution support, the complete PK example, actual generated-code documentation, the centering study's source and evidence, and references. The displayed Julia examples are extracted from the executable feature atlas during rendering.
Prospective guests and cited authors have not reviewed or endorsed the deck.