Transit Compartment Models Versus Delay Differential Equations: When Does Model Choice Matter?

Tóm tắt

This simulation study examined when transit compartment models (TCMs) and delay differential equations (DDEs) for absorption delay lead to materially different conclusions in a one-compartment, first-order, fixed-effects pharmacokinetic setting. While asymptotic theory establishes their equivalence as the number of transit compartments approaches infinity, practical implications for model selection remain incompletely characterized. This simulation study quantifies the convergence rate, statistical distinguishability, and predictive divergence between these frameworks under realistic pharmacokinetic conditions. We found that TCMs approached practical equivalence to DDEs at modest compartment numbers for shorter delays, whereas longer delays required larger . Model distinguishability depended strongly on sampling design: early sampling enabled reliable discrimination, whereas delayed post-absorption sampling provided limited structural discrimination under the adopted likelihood. When both models fit the training data similarly, extrapolation predictions differed by about 14-16% in the simulated scenarios, and repeated cross-validation favored the simpler DDE. Within this one-compartment, first-order, fixed-effects simulation setting, these results support DDE as a parsimonious predictive choice when mechanistic uncertainty remains, and both models fit the observed data adequately.