Why a PBPK Parameter Needs a Record
A PBPK model input is not fully described by a number and a unit. The value \(1.7\ \mathrm{nmol/g}\), for example, could refer to an enzyme per gram of liver tissue or to a different normalization basis; it could come from a direct measurement or a calculation; and it might apply only to a particular population or experimental system. Those differences affect whether the value is interpretable and suitable for a model.
A parameter record makes these distinctions explicit. It is a compact, structured account of what the model variable means, the value supplied to the model, how that value relates to its source, and the conditions under which it can reasonably be used. It is not a claim that the parameter is known without error. Rather, it preserves enough context for another modeler to understand, check, and revisit the input.
A useful record is specific enough to answer three questions: What quantity does the model use? How was the recorded value obtained? For what biological conditions is that value intended to apply? If one of these answers is missing, the number may be impossible to interpret or reproduce.
Fields That Make a Record Interpretable
The fields below provide a practical minimum. A project may add identifiers, version dates, or links to source files, but extra fields do not replace a clear definition of the model variable and its context.
| Field | What to record | Why it matters |
|---|---|---|
| Model variable | The name used in the model and a precise definition of the quantity it represents | Prevents a value being attached to the wrong biological quantity |
| Value and units | The selected value, its units, and, when relevant, the range or distribution used | Lets another user enter and check the input |
| Biological context | Species, population, organ or tissue, age or other relevant characteristics, and experimental or physiological conditions | Defines what system the value describes |
| Source | Enough identifying information to retrieve the report, dataset, or other source | Supports traceability to the evidence used |
| Assay or measurement method | What was measured and by what method, including the material measured where relevant | Clarifies what the reported observation actually represents |
| Transformation history | Each conversion or calculation from reported quantity to model input | Makes derived values auditable |
| Uncertainty | The reported uncertainty, or a clearly labelled modeling range or distribution and its basis | Separates a point input from uncertainty about that input |
| Applicability limits | Known conditions or populations for which transfer is not established or should be treated cautiously | Reduces inappropriate use outside the record’s scope |
A source and an assay are different fields. The source identifies where the information can be found; the assay describes how the relevant observation was generated. Likewise, an assay result and a model parameter are not necessarily the same quantity. A model may require a whole-organ input even when the assay reports a concentration per unit of sample protein.
Keep the Measurement-to-Input Path Visible
A derived model input should retain a short, checkable path from the reported observation to the final value. Each step should preserve the quantity’s meaning and units. Dimensional analysis is a simple way to catch missing or inappropriate conversions: units that cancel should cancel, and the remaining units should match the model variable.
For example, suppose an assay reports enzyme abundance per milligram of microsomal protein, while the model requires abundance per gram of liver. Multiplication by microsomal protein per gram of liver cancels the protein-mass unit and produces enzyme amount per gram of liver. That conversion is valid only if the multiplier is applicable to the biological context and the definitions of the two quantities are compatible.
Worked Example: Recording Hepatic Enzyme Abundance
All values in this example are hypothetical. A PBPK model requires hepatic enzyme abundance per gram of liver. A fictitious assay report describes a mean of \(42\ \mathrm{pmol/mg}\) microsomal protein, with a reported range of \(34\) to \(50\ \mathrm{pmol/mg}\). For illustration, the modeler uses a microsomal protein content of \(40\ \mathrm{mg/g}\) liver as a fixed conversion factor.
Model variable. Hepatic enzyme abundance per gram of liver, denoted here as \(E_{\mathrm{H}}\). The required unit is \(\mathrm{nmol/g}\) liver.
Transformation. Multiply the assay result by the microsomal protein content:
The range transforms in the same way because the conversion factor is held fixed:
Record summary. The model value is \(1.68\ \mathrm{nmol/g}\) liver; the transformed assay range is \(1.36\) to \(2.00\ \mathrm{nmol/g}\). The biological context is the liver samples represented by the hypothetical assay report. The source is the fictitious assay report, and the method is the report’s enzyme quantification assay. The transformation history records multiplication by \(40\ \mathrm{mg/g}\) liver and conversion from picomoles to nanomoles. The uncertainty field should state that the range is the transformed reported range and that it treats the conversion factor as fixed. It does not include uncertainty in that factor. Applicability limits include the population and sample conditions represented by the report; the example provides no basis for extending the value to other populations.
This record is more informative than storing \(1.68\ \mathrm{nmol/g}\) alone: it identifies the normalization basis, exposes the conversion, and makes a limitation of the stated range visible.
Worked Example: Recording Tissue Blood Flow
All values in this example are hypothetical. A model requires total hepatic inflow in \(\mathrm{L/h}\). A fictitious imaging report gives a portal-vein flow of \(0.95\ \mathrm{L/min}\) and a hepatic-artery flow of \(0.30\ \mathrm{L/min}\) for a hypothetical healthy adult under the report’s measurement conditions. For this record, total hepatic inflow is defined as the sum of these two flows.
Model variable. Total hepatic inflow, \(Q_{\mathrm{liver}}\), in \(\mathrm{L/h}\). Stating that it includes both portal-vein and hepatic-artery flow avoids confusing it with either component alone.
Transformation. First sum the components, then convert minutes to hours:
Suppose the report gives a component-based range that yields a total-flow range of \(1.10\) to \(1.40\ \mathrm{L/min}\). Converting the endpoints gives \(66\) to \(84\ \mathrm{L/h}\). The parameter record should identify this as the range reported or derived from the hypothetical report, not as a confidence interval unless the source establishes that interpretation.
Record summary. The value is \(75\ \mathrm{L/h}\), and the model variable is total hepatic inflow, defined as portal-vein plus hepatic-artery flow. The biological context includes the report’s healthy-adult population and its measurement conditions. The source is the fictitious imaging report; the measurement method is the stated flow-imaging procedure. The transformation history records summing the two components and multiplying by \(60\ \mathrm{min/h}\). The uncertainty field records the \(66\) to \(84\ \mathrm{L/h}\) range and its basis. Applicability limits should not imply that the value applies unchanged to other populations or physiological conditions.
For flow, the anatomical definition is as important as the unit. A value in \(\mathrm{L/h}\) does not reveal whether it is total inflow, one vascular component, or flow assigned to a different tissue boundary.
Audit a Record Before Using It
A record can be checked systematically. First, compare the definition and units with the model’s input requirement. Next, follow the transformation arithmetic and verify that the units cancel correctly. Finally, check whether the stated context and limits support the intended use. A numerically plausible value can still be unsuitable if it describes a different quantity or biological system.
Write what the parameter represents, its anatomical or biological basis, and the unit required by the model.
Record where it came from, what was measured, and the method used, without treating the observation as automatically identical to the model input.
Write the arithmetic with units, and check that the final units match the model variable.
Label the uncertainty information accurately and specify the population, tissue, or conditions to which the record pertains.
Worked Example: Catching a Unit and Definition Mismatch
A hypothetical parameter table says only “liver flow = \(1.25\ \mathrm{L/min}\).” The model input is total hepatic inflow in \(\mathrm{L/h}\). The accompanying fictitious measurement note identifies \(1.25\ \mathrm{L/min}\) as the sum of portal-vein and hepatic-artery flow.
The quantity definition is consistent with the model, but the units are not. Convert the value:
The record can therefore specify a value of \(75\ \mathrm{L/h}\), with total hepatic inflow as its definition and the original \(\mathrm{L/min}\) value and conversion preserved in the transformation history. Without the measurement note, the conversion would fix the unit mismatch but not establish that the value represented total inflow. A bare table entry cannot answer that anatomical question.
Common Failure Modes and Practical Choices
- Recording a number without its definition. “Liver flow” does not specify the vascular components included. State the anatomical quantity explicitly.
- Copying a reported value directly into the model. A reported concentration or normalized measurement may not have the units or basis required by the model. Document and check the conversion.
- Listing units without a transformation history. Matching final units do not demonstrate how a derived value was obtained. Preserve the inputs and arithmetic.
- Calling every range a confidence interval. A reported range, a sensitivity range, and a statistical interval have different meanings. Label the range according to what is known about its basis.
- Leaving applicability limits blank when scope is material. A value measured in one population or condition does not automatically establish transfer to another.
A flat parameter table is efficient for entering values, but it often cannot preserve the measurement method, transformation, or scope without becoming difficult to read. A narrative note can explain those details, but it is harder to scan and compare across inputs. A structured record combines the advantages: fields make essential information easy to find, while concise notes preserve details that do not fit a single value cell. No format can compensate for missing source information or justify applying a value beyond its documented scope.
Check Your Understanding
Try each question, then compare your reasoning with the solutions below.
- A model requires enzyme abundance in \(\mathrm{nmol/g}\) liver. An assay reports \(\mathrm{pmol/mg}\) microsomal protein. Name two additional pieces of information needed to convert the assay value to the model input.
- A tissue flow value is reported as \(0.80\ \mathrm{L/min}\), but the model requires \(\mathrm{L/h}\). What is the converted value?
- Why is “hepatic blood flow” an incomplete variable definition when a model input represents total hepatic inflow?
- A record lists a range but does not say whether it is a reported range, statistical interval, or modeling range. What is the documentation problem?
Solutions. (1) The conversion factor for microsomal protein per gram of liver and a clear definition of the assay measurement are needed; the biological context must also support applying that factor. (2) \(0.80\ \mathrm{L/min}\times60\ \mathrm{min/h}=48\ \mathrm{L/h}\). (3) The phrase does not specify which vessels or components are included, so it may not describe the model’s total-flow quantity. (4) The range’s meaning and basis are ambiguous, so its interpretation and use in uncertainty analysis cannot be assessed reliably.