Optimizing Piecewise Infusion Rate Profiles

Why Use a Piecewise Infusion Profile?

A single constant infusion rate is simple to administer and analyze, but it may not provide the desired exposure while also respecting a concentration ceiling. A piecewise-constant profile allows the infusion rate to change at a planned time. The design problem is to calculate the rates for those segments—not merely to choose a total dose—so that the resulting concentration trajectory meets the intended target and remains feasible.

This tutorial considers a fixed two-segment schedule in a one-compartment model with linear elimination. The emphasis is on expressing exposure and concentration limits as functions of the segment rates, solving for rates, and checking whether a target can actually be attained. The rates are decision variables, as in the earlier material on representing dose schedules as optimization variables.

Model and assumptions: Let \(C(t)\) be concentration in mg/L, \(R(t)\) the infusion rate in mg/h, \(V\) the apparent volume in L, and \(k\) the first-order elimination rate constant in h\(^{-1}\). During infusion, concentration follows \(dC/dt=R(t)/V-kC(t)\). Assume \(V\) and \(k\) are constant, the model is linear over the concentrations considered, the initial concentration is known, and the planned rate can be maintained exactly. Here the initial concentration is zero. These assumptions make concentration and exposure linear functions of the infusion rates.

Representing the Schedule and Its Constraints

Divide the 8-hour infusion into two intervals of 4 hours each. Set the rate to \(R_1\) from hour 0 to hour 4, then to \(R_2\) from hour 4 to hour 8. Both rates are measured in mg/h. For a concrete hypothetical design, take \(V=25\) L and \(k=0.2000\) h\(^{-1}\), with rates constrained by \(0\le R_1,R_2\le10\) mg/h. Suppose the concentration must not exceed 1.30 mg/L at any time during the infusion.

On any interval with a constant rate \(R\), starting concentration \(C_0\), and elapsed time \(u\), the model solution is

C(u)=C_0e^{-ku}+\frac{R}{Vk}\left(1-e^{-ku}\right).

Because this trajectory moves monotonically toward the interval’s equilibrium concentration \(R/(Vk)\), its maximum or minimum on that interval occurs at an endpoint. Therefore, for this model and a constant rate within each segment, checking the concentrations at hours 0, 4, and 8 is sufficient to verify the concentration ceiling throughout the infusion.

Deriving Exposure and Concentration Equations

Let \(q=e^{-0.2(4)}=e^{-0.8}\approx0.449329\), and let \(a=1-q\approx0.550671\). The concentration at hour 4 is

C(4)=\frac{a}{Vk}R_1=\frac{0.550671}{5}R_1 \approx0.110134R_1.

Applying the interval solution again from hour 4 to hour 8 gives

C(8)=qC(4)+\frac{a}{Vk}R_2 \approx0.0494865R_1+0.110134R_2.

A convenient way to derive the total exposure from hour 0 to hour 8 is to integrate the differential equation. Rearranging it gives \(C(t)=R(t)/(Vk)-C'(t)/k\). Integrating both sides over the full interval, and using \(C(0)=0\), yields

\mathrm{AUC}_{0-8} =\int_0^8 C(t)\,dt =\frac{4R_1+4R_2}{Vk}-\frac{C(8)}{k}.

Here \(Vk=25(0.2000)=5\), so substituting the endpoint concentration equation gives

\mathrm{AUC}_{0-8} \approx0.552568R_1+0.249329R_2 \quad\text{mg}\cdot\text{h/L}.

The coefficients have units of \((\text{mg}\cdot\text{h/L})/(\text{mg/h})\). They differ because drug infused earlier has more time to be eliminated before hour 8. The second segment contributes to exposure too, but its effect on the endpoint and total exposure differs from that of the first segment.

Formulation: For a specified exposure target \(A^*\), choose \(R_1\) and \(R_2\) to satisfy \(0.552568R_1+0.249329R_2=A^*\), subject to \(0\le R_1,R_2\le10\) mg/h and \(0.0494865R_1+0.110134R_2\le1.30\) mg/L. Also check \(C(4)=0.110134R_1\le1.30\) mg/L. These linear relations define feasibility for this fixed two-segment design.

Worked Example: Finding Rates for a Feasible Exposure Target

Suppose the target is \(7.00\) mg·h/L. Choose \(R_1=10.0\) mg/h and solve the exposure equation for \(R_2\).

7.00=0.552568(10.0)+0.249329R_2, \qquad R_2=\frac{7.00-5.52568}{0.249329} \approx5.913\ \text{mg/h}.

Using the unrounded rate, the calculated exposure is approximately \(7.00\) mg·h/L. The concentration at hour 4 is

C(4)=0.110134(10.0)\approx1.101\ \text{mg/L}.

At hour 8, it is

C(8)\approx0.0494865(10.0)+0.110134(5.913) \approx0.4949+0.6512=1.146\ \text{mg/L}.

Both rates are between 0 and 10 mg/h, and both endpoint concentrations are below 1.30 mg/L. Since concentration is monotone within each constant-rate segment, it also stays below 1.30 mg/L between the checked endpoints. Thus this schedule satisfies the target and the stated constraints.

Attainability Is Part of the Calculation

Solving the exposure equation for one rate does not by itself establish that a schedule is feasible. A calculated rate may violate its allowed range or cause a concentration ceiling to be exceeded. Before choosing a schedule, it is useful to determine whether the constraints permit the requested exposure at all.

For this example, the largest feasible exposure can be found from the linear constraints. The rate \(R_1\) has a larger exposure contribution per unit of endpoint concentration than \(R_2\): the respective ratios are \(0.552568/0.0494865\approx11.17\) and \(0.249329/0.110134\approx2.264\). To maximize exposure under the endpoint concentration ceiling, use the maximum allowed \(R_1=10\) mg/h first, then use the remaining endpoint concentration allowance for \(R_2\). The endpoint constraint gives

R_2\le \frac{1.30-0.0494865(10)}{0.110134} \approx7.310\ \text{mg/h}.

This value is within the rate limit, and \(C(4)\approx1.101\) mg/L is also within the ceiling. The largest exposure is therefore approximately

0.552568(10)+0.249329(7.310) =5.52568+1.82260 \approx7.348\ \text{mg}\cdot\text{h/L}.

The calculation establishes a useful feasibility test: any requested exposure above about 7.348 mg·h/L is unattainable under these assumptions and constraints. Increasing the rates to force the target would violate at least one bound.

Worked Example: Showing That a Target Is Infeasible

Suppose the target is \(7.50\) mg·h/L. Even at the maximum feasible exposure, the schedule reaches only about \(7.348\) mg·h/L. The target therefore cannot be met while retaining all the stated constraints.

The issue is visible if \(R_1=10.0\) mg/h is substituted into the exposure equation and the target is solved without first imposing the concentration ceiling:

R_2=\frac{7.50-0.552568(10.0)}{0.249329} \approx7.919\ \text{mg/h}.

That rate is below the 10 mg/h rate limit, but its predicted endpoint concentration is

C(8)\approx0.0494865(10.0)+0.110134(7.919) \approx0.4949+0.8721=1.367\ \text{mg/L}.

Since 1.367 mg/L exceeds the 1.30 mg/L ceiling, this rate pair is not feasible. The maximum-exposure calculation confirms that changing \(R_1\) cannot recover the target under the same constraints. A different target or a justified change to the design constraints or assumptions would be needed.

Worked Example: Checking a Candidate Schedule Before Adjusting It

Consider the simple schedule \(R_1=8.00\) mg/h and \(R_2=8.00\) mg/h. Its exposure is

\mathrm{AUC}_{0-8} =0.552568(8.00)+0.249329(8.00) =4.420544+1.994632 \approx6.415\ \text{mg}\cdot\text{h/L}.

It falls short of the \(7.00\) mg·h/L target. The concentration checks are

C(4)=0.110134(8.00)\approx0.8811\ \text{mg/L}, \qquad C(8)\approx0.0494865(8.00)+0.110134(8.00) \approx1.277\ \text{mg/L}.

Both rates satisfy their bounds, and both endpoint concentrations are below the ceiling; monotonicity within each segment confirms the ceiling throughout. This candidate is safe under the specified model and limits but does not attain the exposure target. A schedule should be judged against all its requirements, not just its rate or concentration constraints.

Diagnostics, Alternatives, and Limitations

A practical check follows directly from the formulation: calculate the requested exposure from the rates, calculate concentrations at every segment boundary, confirm the rate bounds, and verify the concentration ceiling over each segment. If the target equation gives a negative rate, a rate above its limit, or an excessive boundary concentration, the proposed solution is not feasible. If the model predicts an interior concentration extremum, or if rates change continuously rather than at the planned boundaries, the endpoint-only check needs to be reconsidered.

A constant-rate infusion is the simplest alternative. It has only one rate to choose, but it cannot independently shape early and late input. A two-segment schedule offers an additional degree of control while retaining transparent calculations. More segments can represent more flexible profiles, but require more decision variables and reliable specification of the change times. Numerical optimization can also handle more complex models; it does not remove the need to verify model assumptions, constraints, and units.

The conclusions here are conditional on a linear one-compartment model with known parameters and a known initial concentration. Parameter uncertainty, changes in clearance, nonlinear elimination, delays in drug delivery, and administration errors can all change the realized exposure or concentration profile. The calculated rates are therefore model-based design values, not guarantees about an individual’s observed concentrations. A model that poorly describes the intended setting may produce a mathematically feasible but clinically inappropriate schedule.

Common Pitfall

Wrong reasoning: “The rate obtained by solving the exposure equation is below the maximum rate, so the target is feasible.”
Correction: Rate limits are only part of feasibility. In the \(7.50\) mg·h/L example, the calculated \(R_2\approx7.919\) mg/h is below 10 mg/h, yet it implies \(C(8)\approx1.367\) mg/L and violates the 1.30 mg/L ceiling. Check every constraint and, when needed, calculate the maximum feasible exposure.
Arithmetic check: Keep the rate-specific exposure contributions separate before adding them. For instance, \(0.552568(8)=4.420544\), not 6.415; the sum for the two 8 mg/h rates is about 6.415 mg·h/L.

Key takeaway: A piecewise infusion design is a constrained calculation over segment rates. Derive the model-based exposure and concentration equations, solve for candidate rates, and verify every rate and concentration bound before calling the target attainable.

Check Your Understanding

Answer each question using the two-segment model and limits developed above.

For \(R_1=6.00\) mg/h and \(R_2=6.00\) mg/h, calculate the predicted exposure. Does this schedule meet a target of \(7.00\) mg·h/L?
Why is checking concentrations at hours 0, 4, and 8 sufficient for this two-segment model?
Can a target of \(7.40\) mg·h/L be attained under the stated constraints? Explain using the maximum feasible exposure.
For the feasible \(7.00\) mg·h/L schedule in the worked example, name the rate and concentration checks that establish feasibility.

Solutions. (1) The exposure is \((0.552568+0.249329)(6.00)=4.811\) mg·h/L, so it falls short of 7.00 mg·h/L. (2) Within each constant-rate segment, concentration moves monotonically toward its equilibrium, so its extrema on that segment occur at its endpoints. (3) No. The maximum feasible exposure is about 7.348 mg·h/L, below 7.40 mg·h/L. (4) \(R_1=10.0\) and \(R_2\approx5.913\) mg/h are each within 0–10 mg/h; \(C(4)\approx1.101\) and \(C(8)\approx1.146\) mg/L are below 1.30 mg/L, and monotonicity verifies the ceiling between endpoints.