The colon is divided into 20 equal axial regions. Distal end is an approximate rectal proxy, not a separate anatomically validated rectum. Inflammation settings are illustrative; affected segments have twice the baseline paracellular flux and 1.5 times lamina-propria inactivation, not measured disease effects. There is no epithelial or mucus damage model. The capsule and upper GI are not modeled: the user-specified active fraction is an assumed lumped arrival fraction. Release is first order and the effect of fiber and microbiota is not explicitly modeled. Luminal carrier and free protein move distally. Each region includes mucus, epithelial endosomes, epithelial cytosol and lamina propria. Colonic lymphoid follicles, draining lymph nodes and plasma are lumped compartments. No target production or turnover is represented; the generic binding sink is not a prediction of soluble TNF neutralization. Values are active protein mass, not concentration. Loss includes inactivation and outflow.

No formulation-specific coefficients are known. Size dependence is an illustrative mathematical assumption, not a validated property. CPP multiplies endosomal uptake only; escape is a separate adjustable rate and transcytosis is not inferred from uptake. Enhancer effects are generic, not matched to any named ingredient or dose. Each dose supplies an arbitrary co-localized signal that moves with luminal flow and decays exponentially; it temporarily multiplies either paracellular transport or endosomal transcytosis. The effect half-life is an assumption, not an estimate of duration of barrier changes in a person. CPP and enhancer effects are independently specified and have not been validated in combination. Follicle entry, lymphatic drainage, blood entry and target binding rates are hypothetical. The model excludes immune responses, anatomy-specific gradients within tissue, receptor kinetics and clinical effects.