Common Errors in Segmental Transit Models

Why Segmental Models Can Go Wrong

A segmental transit model may produce smooth concentration-time curves while still representing the wrong mass flows. Errors can arise when an amount is used as though it were a concentration, when a flux is counted twice, or when the index on a transit rate does not refer to the segment the drug leaves. A further error is to treat drug still present in the gut at the simulation end as though it had already been absorbed or eliminated.

These problems are best diagnosed by writing down the meaning and units of every state and flux before interpreting a simulation. The equations below use a simple chain of \(N\) gut segments. They are a bookkeeping framework for diagnosing implementation errors, not a claim that every model must use these particular kinetics.

For this framework, solid drug dissolves in its current segment but does not itself transit; dissolved drug may transit, be absorbed, or remain in the segment. An input \(I_i\), if present, is an amount rate entering the dissolved state of segment \(i\). Transit from segments \(1,\ldots,N-1\) enters the next segment. \(J_N\) leaves the modeled gut chain, so it is counted as gut exit, not absorption. These assumptions need to match the model being implemented: if solid particles also transit, their states and transfers require their own explicit balances.

The terminal transit term \(J_N\) is an outflow from the last segment. It must either be included as a gut-exit flux, as above, or be replaced by a separately specified downstream compartment. It must not disappear from the accounting merely because there is no segment \(N+1\) in the model.

Start With Units and State Meaning

An amount and a concentration are not interchangeable. If \(A_i\) is in mg, a first-order absorption flux \(U_i=k_{\mathrm{a},i}A_i\) is in mg/h. If the model instead uses concentration, an absorption clearance \(CL_{\mathrm{a},i}\) in L/h gives the amount flux \(U_i=CL_{\mathrm{a},i}C_i\), also in mg/h. For a fixed volume, the two formulations agree when \(CL_{\mathrm{a},i}=k_{\mathrm{a},i}V_i\). The parameter units therefore depend on the state representation.

A reliable review begins by writing the units beside each state, parameter, and rate. Every term added to or subtracted from a state derivative must have units of amount per time. A concentration derivative instead has units of concentration per time, so amount fluxes cannot be inserted into it without the appropriate volume conversion. If volumes change with time, converting between amount and concentration also requires accounting for that change; simply dividing an amount derivative by the current volume may not be sufficient.

Count Every Transfer Once

A transfer from one state to another has two appearances in the state equations: it is subtracted from the source and added to the destination. These are not two transfers; together they describe one movement of mass. Likewise, dissolution is one flux from solid to dissolved drug: it depletes \(S_i\) and supplies \(A_i\). If a model has an explicit dissolution flux \(D_i\), adding a second dissolution contribution to \(A_i\) without a distinct physical process duplicates drug.

The same rule applies to transit. For \(i<N\), the flux \(J_i\) appears as \(-J_i\) in the balance for \(A_i\) and \(+J_i\) in the balance for \(A_{i+1}\). If the receiving state gets the flux twice, or if the source loses it twice, the total modeled mass changes incorrectly.

A Mass-Balance Derivation

Let \(R_{\mathrm{abs}}\) be cumulative absorbed amount and \(R_{\mathrm{exit}}\) cumulative amount exiting the last gut segment. Their rates are \(dR_{\mathrm{abs}}/dt=\sum_{i=1}^N U_i\) and \(dR_{\mathrm{exit}}/dt=J_N\). The total accounted amount is

To verify conservation, differentiate \(M(t)\). Each solid balance contributes \(-D_i\), and each dissolved balance contributes \(+D_i\), so dissolution cancels segment by segment. For every internal transit flux \(J_i\), \(i<N\), the source contributes \(-J_i\) and the next segment contributes \(+J_i\). Absorption losses \(-U_i\) cancel the corresponding increases in \(R_{\mathrm{abs}}\). The terminal loss \(-J_N\) cancels the increase in \(R_{\mathrm{exit}}\). Thus, if all inputs are zero, \(dM/dt=0\). With inputs, the change in \(M\) must equal the total input.

This derivation also gives a practical diagnostic: compute the residual between the change in total accounted amount and the integrated input. It should be close to zero apart from numerical integration error. A steadily growing residual is not explained by a plausible absorption profile; it points to inconsistent bookkeeping or a numerical implementation problem.

Check the Segment Index

In a sequential chain, the index on a transit flux should identify where the drug leaves. With \(J_i=k_{\mathrm{tr},i}A_i\), the outflow from segment \(i\) is a loss from \(A_i\) and, for \(i<N\), an inflow to \(A_{i+1}\). This convention prevents a frequent off-by-one error: using the receiving segment’s transit rate to calculate the incoming flux.

Do Not Discard Drug Still Downstream

A simulation reports states at its chosen stopping time, not necessarily the ultimate fate of every dose. At that time, drug may remain as solid or dissolved amount in one or more gut segments. The observed cumulative absorption up to that time is not automatically the total amount that will eventually be absorbed. Likewise, drug that has not yet left the modeled gut is not gut exit.

To compare eventual absorption across simulations, either simulate long enough that relevant residual states are negligible, extend the model to track downstream drug, or explicitly account for remaining drug using a justified continuation of the model. Merely adding all residual gut amount to cumulative absorption is incorrect: some residual amount may later exit the gut or remain undissolved under the model assumptions.

Diagnostics, Alternatives, and Limitations

Several checks complement inspection of the concentration-time profile. First, verify dimensions for every equation term. Second, inspect each transfer in both its source and destination balances. Third, compare source outflows and destination inflows for every internal link. Fourth, calculate the total-mass residual over time and inspect residual segment amounts at the simulation end. A sudden residual jump often points to an incorrect flux or index; a gradual drift may indicate repeated small errors or numerical integration error.

Amount-state and concentration-state formulations are both valid when implemented consistently. Amount states make conservation accounting direct; concentration states can be convenient when rates are naturally expressed using concentrations and clearances. They should not be mixed casually: conversion requires the relevant volume and consistent treatment of volume changes. Similarly, cumulative absorbed and exit amounts can be represented as explicit model states or calculated by integrating their fluxes after simulation. Either approach is valid if each flux is counted once and the reported totals use the same time horizon as the simulated states.

Mass balance is a powerful diagnostic but does not establish that the chosen transit structure or parameter values adequately describe physiology. A model can conserve mass perfectly and still use inappropriate assumptions, inputs, or rates. Conversely, a small numerical balance residual should be judged relative to solver tolerances and the scale of the dose. If residual amounts are substantial at the endpoint, the simulation duration is inadequate for an interpretation of eventual disposition, even when the numerical solution itself is accurate.