Several prominent commercial software packages explicitly fix or strictly constrain Vp and Q during individual data fitting. In clinical therapeutic drug monitoring (TDM), this practice is known as fixing parameters or modeling with zero inter-individual variability (IIV) on peripheral parameters.
Modern Model-Informed Precision Dosing (MIPD) platforms handle the parameters as follows:
| Software Package | How Vp and Q Are Handled | Structural Approach |
|---|---|---|
| InsightRX Nova | Fixed / Limited fitting (Model Dependent) | Pop-PK models such as Goti et al. or Thomson are hardcoded into the software. Many of these foundational models mathematically fixed the IIV of Q and Vp to 0% during the initial model building phase to ensure stability. |
| PrecisePK | Fixed / Constrained | Primarily uses established two-compartment literature models. If a clinician enters only one serum concentration (e.g., a trough), the math lacks the "identifiability" to change 4 distinct parameters. The algorithm locks Vp and Q to the population average and shifts only Vc and CL. |
| BestDose (LAPK) | Constrained / Not fixed | Unlike others, BestDose utilizes a Non-Parametric Bayesian approach. It allows a full grid of parameter variations, but it heavily weights the adjustments toward CL and Vc if clinical data points are sparse. |
Commercial software forces this restriction for three critical reasons:
In real-world hospital workflows, pharmacists typically check only one trough level, or occasionally a peak and a trough. [1]
Clinicians use software to achieve a target AUC24 and avoid acute kidney injury. Because AUC = Dose / CL, CL is the only parameter that dictates steady-state exposure. Forcing the software to focus its mathematical energy on optimizing CL and Vc guarantees that the calculated AUC is as accurate as possible, even if the precise internal tissue distribution remains a minor mystery. [3] [4]
When software vendors build a two-compartment model, they must use a published Population Pharmacokinetic (Pop-PK) dataset. When researchers first created these models in software like NONMEM or Monolix, they frequently discovered that trying to calculate a unique Vp or Q variance for every patient added data "noise" without improving predictive power. They intentionally set the Inter-Individual Variability (ω²) of Vp and Q to exactly zero, a setting that commercial platforms carry forward. [5] [6]
Links open in a separate, closable popup window. Ctrl/Cmd-click, middle-click, or right-click → "Open link in new tab" to open a reference in a new tab instead.