Tutorials › Pharmacometrics › Communicating QSP Model Results
Pharmacokinetics · Quantitative Systems Pharmacology

Communicating QSP Model Results

Learn how to turn quantitative systems pharmacology simulations into clear scientific evidence by separating model predictions from observations, showing uncertainty and sensitivity, comparing scenarios transparently, and connecting model outputs to the biological and clinical questions that matter.

Intermediate QSP Modeling Model Interpretation Pharmacometrics
01 · The big picture

1. Why Communicating QSP Results Is a Modeling Skill

A quantitative systems pharmacology (QSP) model can contain hundreds of variables, parameters, equations, biological mechanisms, and simulated scenarios. Yet the purpose of presenting the model is rarely to show all of that information at once. The purpose is to communicate what the model contributes to a scientific question.

Good QSP communication therefore requires more than producing attractive plots. The analyst must explain what was simulated, what the model predicts, how certain those predictions are, which assumptions matter, and what conclusions are justified by the analysis.

QSP model mechanisms parameters · data Simulation scenarios predictions Scientific story evidence · uncertainty interpretation The communication layer connects model output to a scientific decision or question.

A useful QSP presentation moves from model structure to simulation results and finally to a clearly stated scientific interpretation.

Core idea: the goal is not to communicate everything the model contains. The goal is to communicate the model-based evidence that is relevant to the scientific question, together with enough context to understand its uncertainty and limitations.
02 · Know the audience

2. Start With the Scientific Audience

The same QSP analysis may need to be communicated to modelers, pharmacologists, clinical scientists, statisticians, regulatory scientists, or project teams. Each audience needs a different level of technical detail.

Audience Usually needs to understand Technical detail to emphasize
QSP modelers Mechanistic assumptions, equations, parameterization, identifiability, and diagnostics Model structure, parameter uncertainty, sensitivity, calibration, and validation
Clinical scientists What the model predicts for disease, treatment, and patient-relevant outcomes Biological interpretation, treatment scenarios, response distributions, and limitations
Project teams How the model changes understanding of a development question Scenario comparisons, key drivers, uncertainty, and implications
Regulatory audiences Whether the evidence is scientifically justified and appropriately qualified Model purpose, assumptions, verification, validation, uncertainty, and applicability

A common communication failure is to begin with the model rather than the question. A presentation can spend ten slides explaining compartments and equations before the audience learns what decision the simulation was intended to inform.

Practical rule: begin with the scientific question, not with the number of compartments, equations, or parameters in the model.
03 · Frame the question

3. State the Question Before Showing the Result

A QSP result is meaningful only in relation to a question. Before presenting a plot, state what the simulation is intended to investigate.

Examples include:

  • How does target inhibition change downstream biomarker dynamics?
  • What degree of target engagement is required to produce a specified biological effect?
  • Which mechanisms could explain a difference in treatment response?
  • How might changes in dose or schedule affect the predicted response?
  • Which biological parameters contribute most strongly to uncertainty in the outcome?
  • What experimental observations would most reduce uncertainty in the model?

These questions lead to different analyses. A dose-response question may emphasize scenario simulations, whereas an uncertainty question may emphasize parameter distributions and sensitivity analysis.

\[ \text{Scientific question} \rightarrow \text{Model quantity} \rightarrow \text{Simulation} \rightarrow \text{Evidence} \rightarrow \text{Interpretation} \]

This sequence provides a simple organizing principle for both figures and presentations.

04 · Prediction versus observation

4. Clearly Separate Observed Data From Model Predictions

One of the most important principles in communicating mechanistic model results is to distinguish what was actually observed from what was generated by the model.

Information type Example How to describe it
Observed Measured biomarker concentration "Observed concentration"
Model-predicted Simulated biomarker trajectory "Model-predicted trajectory"
Calibrated Parameter estimated from experimental data "Estimated/calibrated parameter"
Assumed Fixed biological parameter based on literature "Literature-based fixed value"
Scenario-derived Predicted response under an untested dose "Simulated scenario"

The distinction becomes particularly important when a model is used to extrapolate beyond the experimental conditions used for calibration. A simulated result can be scientifically useful without being an observation.

Communication principle: never make a model prediction visually or verbally indistinguishable from an experimental observation.
05 · Choose the output

5. Choose Outputs That Answer the Question

A complex QSP model may generate hundreds or thousands of state variables. Most of these should not appear in the primary communication of results.

The first task is therefore to identify the decision-relevant output. This may be a biomarker, target engagement measure, disease burden, response probability, exposure metric, or other endpoint.

Question Potential output Useful visualization
What happens over time? Biomarker or disease-state trajectory Time-course plot
How does response depend on dose? Endpoint versus dose Dose-response curve
Which mechanism drives the result? Output sensitivity to parameters Tornado plot or sensitivity plot
How uncertain is the prediction? Prediction distribution Prediction interval or uncertainty band
How do treatments compare? Difference or ratio between scenarios Scenario comparison plot
Which biological pathway changes? Multiple mechanistic outputs Pathway or mechanism diagram

A useful output is not necessarily the variable that is most interesting from a modeling perspective. It is the variable that best connects the model to the scientific question.

06 · Time-course results

6. Communicating Dynamic QSP Predictions

Many QSP models are explicitly dynamic. They describe how concentrations, signaling states, cell populations, disease variables, and other quantities change over time. Time-course plots are therefore among the most common QSP visualizations.

0 Time Output Model prediction Prediction alternative Observed data

A time-course figure should make the distinction between observations and model-generated trajectories immediately apparent.

Time-course plots should identify the units of both axes, the treatment or scenario, and whether lines represent individual simulations, median predictions, means, or other summaries.

If uncertainty is available, it is often more informative to show an uncertainty band than to display a single deterministic trajectory.

07 · Uncertainty

7. Communicating Model Uncertainty

QSP predictions are rarely known exactly. Uncertainty can arise from parameter estimation, biological variability, measurement error, structural assumptions, or uncertainty about mechanisms.

For a predicted quantity \(Y\), uncertainty can be represented conceptually as a distribution rather than a single number:

\[ Y_{\text{pred}} \sim p(Y\mid \text{data},\text{model},\text{assumptions}) \]

The exact interpretation of a prediction interval depends on how the simulation was constructed. What matters for communication is that the audience can distinguish the central prediction from the range of plausible outcomes represented by the analysis.

Presentation What it communicates Potential limitation
Single trajectory Central model prediction Can hide substantial uncertainty
Confidence or credible interval Parameter or inference uncertainty under a specified framework Interpretation depends on the statistical framework
Prediction interval Range of future or simulated outcomes Must be defined clearly
Simulation envelope Range of model-generated trajectories Can be misinterpreted if simulation assumptions are unclear
Individual trajectories Between-subject or scenario variability Can become visually crowded
Important: an uncertainty band is only useful when the audience knows what generated it. Label whether the band represents parameter uncertainty, between-subject variability, residual variability, scenario variation, or another quantity.
08 · Sensitivity

8. Communicating Sensitivity Analysis

Sensitivity analysis asks how strongly a model output changes when model inputs or parameters change. It can help identify which biological assumptions matter most for a particular prediction.

For an output \(Y\) and parameter \(\theta_i\), a local sensitivity can be represented conceptually as:

\[ S_i=\frac{\partial Y}{\partial \theta_i} \]

Global sensitivity analyses may instead explore parameter distributions over a broad range of plausible values.

Lower output Higher output Target turnover Binding affinity Production rate Clearance Baseline level

A sensitivity plot is most informative when the output, parameter ranges, and direction of influence are explicitly defined.

A sensitivity ranking should not automatically be interpreted as a ranking of biological importance. A parameter may appear influential because its plausible range is wide, because of model parameterization, or because of correlations with other parameters.

Better interpretation: say that a parameter is influential for the specified output under the specified analysis conditions rather than declaring it universally important.
09 · Scenario comparisons

9. Communicating Scenario Simulations

Scenario simulation is one of the major strengths of QSP modeling. Once the model has been calibrated and evaluated, investigators can explore conditions that may not yet have been experimentally tested.

Examples include:

  • Different dose levels.
  • Different dosing intervals.
  • Different degrees of target engagement.
  • Changes in biological pathway activity.
  • Alternative mechanisms of action.
  • Different patient or disease characteristics.
  • Combination-treatment scenarios.

A scenario comparison should clearly identify the baseline or reference scenario. Without a reference, it can be difficult to understand whether an apparent difference is scientifically meaningful.

\[ \Delta Y = Y_{\text{scenario}}-Y_{\text{reference}} \]

For some outcomes, a relative comparison may be more useful:

\[ R=\frac{Y_{\text{scenario}}}{Y_{\text{reference}}} \]

The choice between absolute difference, relative difference, ratio, or another measure should follow the scientific question rather than the convenience of the visualization.

10 · Mechanistic interpretation

10. Show the Mechanism Behind the Prediction

QSP models are valuable partly because they provide mechanistic explanations for predictions. A strong presentation therefore does not stop at saying that one scenario produced a larger response.

Instead, explain the causal or mechanistic sequence represented by the model.

Dose exposure Target engagement Pathway modulation Effect response The mechanistic chain explains why the simulated output changes.

Mechanistic communication connects the intervention to intermediate biological processes and ultimately to the endpoint of interest.

This is particularly useful when multiple pathways contribute to an outcome. Showing intermediate variables can make the model's mechanistic interpretation more transparent.

11 · Model evaluation

11. Communicating Model Evaluation and Validation

Before using a QSP model to make predictions, audiences need to understand whether the model has been evaluated against relevant evidence.

Model evaluation can include comparison with experimental observations, historical data, external datasets, known biological behavior, or other independent evidence.

Question What to communicate
Does the model reproduce calibration data? Show observed data and model predictions under calibration conditions.
Does it reproduce independent observations? Show external or validation data separately from calibration data.
Does it reproduce qualitative behavior? Describe the biological pattern and whether the model reproduces it.
Are parameters biologically plausible? Provide ranges, sources, and rationale where appropriate.
Are predictions robust? Show uncertainty and sensitivity analyses.

A visually impressive fit is not, by itself, evidence that the model is suitable for every future prediction. The relevance of the validation evidence depends on the intended use of the model.

Model-use principle: communicate validation in relation to the intended application. Evidence supporting one use of a model does not automatically establish its adequacy for every extrapolation.
12 · Limitations

12. Communicate Limitations Explicitly

A QSP model is a structured representation of biological knowledge. It is therefore useful to communicate not only what the model predicts but also what assumptions constrain those predictions.

Important limitations may include:

  • Limited or sparse experimental data.
  • Uncertain biological parameters.
  • Alternative plausible mechanisms.
  • Structural assumptions in the model.
  • Uncertainty in disease biology.
  • Limited representation of patient heterogeneity.
  • Extrapolation beyond the conditions used for model evaluation.
  • Correlation or non-identifiability among parameters.
  • Uncertainty in the relationship between a model endpoint and a clinical outcome.

A limitation should be described specifically enough that the audience understands how it affects interpretation.

For example, saying "the model has limitations" is much less informative than saying that "the predicted response at high target engagement is sensitive to an uncertain parameter describing pathway turnover."

13 · Sources of uncertainty

13. Distinguish Different Sources of Uncertainty

Not all uncertainty in a QSP analysis has the same meaning. Separating sources of uncertainty can substantially improve scientific communication.

Source Description Typical communication approach
Parameter uncertainty Uncertainty about numerical parameter values Parameter distributions, confidence or credible intervals, sensitivity analysis
Biological variability Real differences among individuals or systems Population simulations or distributions of predicted outcomes
Measurement uncertainty Imprecision in experimental observations Observation-error representation and data uncertainty
Structural uncertainty Uncertainty about model equations or mechanisms Alternative model structures or scenario analyses
Extrapolation uncertainty Uncertainty introduced when predictions extend beyond supporting data Applicability analysis and explicit qualification of predictions
Key distinction: variability describes differences that may genuinely exist in the system, whereas uncertainty describes what is not known precisely. They should not automatically be represented or interpreted in the same way.
14 · Figures

14. Designing Effective QSP Figures

Figures often carry most of the information in a QSP presentation. A good figure should allow the audience to identify the scientific message without reconstructing the analysis themselves.

Useful design principles include:

  • Give every axis a meaningful label and unit.
  • Identify treatment, dose, and scenario definitions.
  • Distinguish observations from predictions.
  • Show uncertainty when it materially affects interpretation.
  • Avoid displaying every simulated variable in a single figure.
  • Use consistent definitions across related figures.
  • State what the shaded region, line, point, or interval represents.
  • Include a reference scenario when making comparisons.

The title can also communicate the scientific question. For example, "Predicted biomarker suppression versus dose" is generally more informative than simply "Simulation Results."

15 · Tables

15. When a Table Is Better Than a Plot

Not every QSP result needs to be visualized as a curve. Tables can be particularly useful for summarizing scenario definitions, parameter assumptions, or key numerical predictions.

Scenario Dose Target engagement Predicted response
Reference Low 35% 0.82
Scenario A Medium 68% 0.61
Scenario B High 88% 0.48

A table is especially useful when exact numerical values matter or when the audience needs to compare several scenarios simultaneously.

16 · Worked example

16. Worked Example: Communicating a QSP Dose-Response Simulation

Consider a hypothetical QSP model describing a drug that inhibits a biological target and produces a downstream disease-related response. Suppose the model is used to simulate three dose levels.

Scenario Predicted target engagement Predicted response
Placebo/reference 0% 1.00
Low dose 45% 0.82
Medium dose 70% 0.63
High dose 88% 0.51

Step 1: Define the comparison

The reference scenario is the untreated condition with an outcome of 1.00. The predicted effect of treatment can therefore be summarized relative to this reference.

Step 2: Calculate the relative change

For the medium-dose scenario:

\[ \text{Relative change} = \frac{0.63-1.00}{1.00} = -0.37 \]

The model therefore predicts a 37% reduction in the outcome relative to the reference scenario.

Step 3: Connect the prediction to mechanism

The model predicts that the medium dose produces approximately 70% target engagement. The reduction in the downstream outcome is therefore interpreted through the modeled relationship between target engagement and the disease-related pathway.

Step 4: Add uncertainty

Suppose the simulation produces a 90% prediction interval of 0.52 to 0.75 for the medium-dose response. The result should then be communicated as a central prediction together with the range represented by the simulation, rather than simply as 0.63.

Step 5: State the scientific interpretation

Example interpretation: Under the assumptions and parameter distributions used in the QSP model, the medium-dose scenario produces substantial predicted target engagement and a lower downstream disease-related output than the reference scenario. The magnitude of the prediction remains dependent on the modeled target-response relationship and associated parameter uncertainty.

Notice what this statement does not say. It does not claim that the simulation proves the treatment will produce the predicted clinical outcome. Instead, it identifies the model-based result, the mechanism represented by the model, and the conditions under which the prediction should be interpreted.

17 · Comparison

17. Communicating Differences Between Scenarios

When comparing scenarios, the most useful result is often the difference between them rather than the absolute prediction from each scenario.

For two scenarios \(A\) and \(B\), the absolute contrast is:

\[ \Delta = Y_A-Y_B \]

A relative contrast can be expressed as:

\[ \text{Relative difference} = \frac{Y_A-Y_B}{Y_B} \]

The appropriate comparison depends on the endpoint. For a bounded biomarker, percentage change may be intuitive. For an event probability or survival-related quantity, another scale may be more appropriate.

Most importantly, define the direction of the comparison. A statement such as "Scenario A is 20% lower" is incomplete unless the reference quantity is clear.

18 · Population predictions

18. Communicating Heterogeneity and Population Simulations

QSP models can be used to explore how differences among simulated individuals affect predicted outcomes. This is particularly useful when biological variability is scientifically important.

Instead of presenting a single deterministic trajectory, a population simulation may produce a distribution:

\[ Y_1,Y_2,\ldots,Y_N \sim \text{Population simulation} \]

The resulting distribution can show a median response, percentile range, probability of exceeding a threshold, or other clinically meaningful summary.

However, simulated heterogeneity should not automatically be interpreted as observed clinical variability. The distribution is conditional on the assumptions used to generate the simulated population.

Communication rule: describe a simulated population distribution as a model-based prediction unless it has been directly validated against appropriate population data.
19 · Thresholds

19. Communicating Threshold-Based Predictions

QSP analyses often investigate whether a biological or clinical threshold is reached. Examples include target engagement above a specified level or disease burden below a specified value.

If \(Y\) is an output and \(T\) is a threshold, the quantity of interest may be:

\[ P(Y\leq T) \]

For a simulated population, this can be estimated as:

\[ \widehat{P}(Y\leq T) = \frac{1}{N} \sum_{i=1}^{N} I(Y_i\leq T) \]

When communicating such a result, report the threshold, the population or simulation conditions, and the definition of the outcome.

20 · Negative results

20. Communicating a Result When the Model Predicts Little or No Effect

A QSP model may predict that changing a parameter, dose, mechanism, or intervention produces little change in the endpoint of interest. This can be scientifically useful.

The interpretation should distinguish between several possibilities:

  • The mechanism genuinely has little influence on the modeled endpoint.
  • The selected parameter range does not span a biologically important regime.
  • Another pathway compensates for the perturbation.
  • The endpoint is insensitive to the mechanism under the simulated conditions.
  • The result is obscured by parameter or structural uncertainty.

A flat response should therefore prompt mechanistic investigation rather than being automatically described as proof that a biological mechanism is unimportant.

21 · Model space

21. Communicating Extrapolation Beyond the Data

One of the defining uses of QSP is to explore conditions that have not yet been experimentally observed. Such predictions can be useful precisely because they extend beyond existing data, but that extension should be explicit.

Region supported by observations Model extrapolation Observed range Extrapolated range

A QSP prediction can extend beyond observed data, but the distinction between data-supported behavior and extrapolation should be visible in the communication.

Extrapolation is not inherently inappropriate. It is one of the reasons mechanistic models are useful. The key is to communicate which portions of the prediction are directly supported by data and which depend more strongly on model assumptions.

22 · Presentation structure

22. A Practical Structure for a QSP Results Presentation

A clear QSP results presentation can often follow a simple progression.

  1. Scientific question. State the decision or biological question.
  2. Model purpose. Explain what the model is intended to represent.
  3. Relevant evidence. Summarize the data and knowledge supporting the model.
  4. Model evaluation. Show how the model reproduces relevant observations.
  5. Primary prediction. Show the main result directly.
  6. Mechanistic explanation. Explain why the model predicts the result.
  7. Uncertainty. Show how much the prediction varies.
  8. Sensitivity. Identify assumptions or parameters that drive the prediction.
  9. Scenario comparison. Explain how alternative conditions change the result.
  10. Limitations. State what the analysis does not establish.
  11. Conclusion. Give the model-based scientific interpretation in one or two sentences.
One-slide test: if someone sees only the primary results slide, they should be able to identify the question, scenario, endpoint, direction of effect, and major uncertainty without needing the model code.
23 · Reporting language

23. Use Precise Model-Based Language

The language used to describe QSP results should reflect the evidentiary status of the result.

Less precise More precise
"The drug reduces disease." "The model predicts a reduction in the disease-related endpoint under the simulated treatment conditions."
"The target causes the response." "The model attributes the predicted response change to the modeled target-mediated pathway."
"The treatment will work." "The simulation predicts the specified outcome under the model assumptions."
"Parameter X is the most important." "The selected sensitivity analysis identifies parameter X as highly influential for this output."
"The model proves the mechanism." "The model provides a mechanistic explanation consistent with the available evidence."

This distinction is not merely linguistic. Precise wording prevents model-generated evidence from being mistaken for direct experimental or clinical evidence.

24 · Reproducibility

24. Make the Simulation Reproducible

A communicated result should be sufficiently documented that another analyst can understand how it was generated.

Important details may include:

  • Model version or release identifier.
  • Parameter set or parameter distributions.
  • Initial conditions.
  • Dosing and intervention assumptions.
  • Simulation duration and time step.
  • Population-generation procedure.
  • Random seed when stochastic simulation is used.
  • Scenario definitions.
  • Output definitions and transformations.
  • Uncertainty and sensitivity-analysis methods.

These details do not all need to appear on the primary results slide. They can be provided in supplementary material, appendices, or technical documentation.

Good practice: make the main communication simple without making the underlying analysis irreproducible.
25 · Common mistakes

25. Common Mistakes When Communicating QSP Results

Mistake Why it causes problems Better approach
Showing every model output Important findings become difficult to identify Select outputs tied directly to the scientific question
Hiding uncertainty Creates an impression of false precision Show relevant prediction or uncertainty ranges
Mixing observations and simulations Blurs the evidentiary distinction Use explicit visual and textual labels
Reporting a sensitivity ranking without context Can be interpreted as universal biological importance Specify output, parameter ranges, and analysis method
Presenting extrapolation as established fact Hides dependence on model assumptions Identify extrapolated conditions explicitly
Overloading figures with model detail The scientific message becomes difficult to see Move technical detail to supporting material
Using causal language too strongly Can overstate what the model establishes Use model-based and evidence-based language
26 · Decision relevance

26. Connect the Result to the Decision

The final step in communicating a QSP result is to explain why the result matters. This does not mean turning a model prediction into a definitive decision. Instead, it means describing how the evidence informs the question that motivated the analysis.

For example, a simulation may indicate that:

  • a particular dose range produces the desired modeled target engagement;
  • additional target engagement produces diminishing incremental response;
  • a prediction is highly sensitive to a particular biological parameter;
  • an alternative mechanism produces substantially different outcomes;
  • additional experimental data would meaningfully reduce uncertainty; or
  • a proposed scenario falls outside the region where the model has strong supporting evidence.

These are useful scientific conclusions because they identify what the model adds to the evidence base without pretending that simulation alone resolves every uncertainty.

27 · Putting it together

27. From a Complex Simulation to a Clear Scientific Message

Imagine that a QSP model contains 250 state variables, 180 parameters, several feedback loops, and simulations of multiple dose levels. The complete model may be technically complex, but the primary result may be summarized by a small number of questions.

Communication question Example answer
What was asked? How does target inhibition affect the downstream disease biomarker?
What was simulated? Three dose levels spanning low to high target engagement.
What was predicted? Higher target engagement produced progressively lower modeled biomarker levels.
Why? The model links target inhibition to suppression of a downstream biological pathway.
How certain? The magnitude of the prediction varies across plausible parameter values.
What drives uncertainty? The result is particularly sensitive to pathway turnover parameters.
What remains uncertain? The model has limited evidence supporting extrapolation at very high target engagement.
What does the model contribute? It provides a mechanistic framework for connecting target engagement to the downstream response.
The communication hierarchy: question → prediction → mechanism → uncertainty → limitation → implication. This sequence often communicates a complex QSP analysis more effectively than a model-centered presentation.
28 · Checklist

28. QSP Results Communication Checklist

Before finalizing a report, presentation, figure, or manuscript section, ask:

  • Is the scientific question stated clearly?
  • Is the purpose of the model explicit?
  • Are observed data clearly distinguished from model predictions?
  • Are treatment and scenario definitions unambiguous?
  • Are the primary outputs directly relevant to the question?
  • Are units and scales clearly labeled?
  • Is uncertainty shown or described when it materially affects interpretation?
  • Is the source of uncertainty identified?
  • Are sensitivity results interpreted within the context of the analysis?
  • Are mechanistic pathways explained rather than merely asserted?
  • Are extrapolations identified?
  • Are major model assumptions stated?
  • Are limitations communicated explicitly?
  • Can the simulation be reproduced from the documented inputs and settings?
  • Does the conclusion distinguish model-based evidence from direct experimental evidence?

29. Key Takeaways

  • Start with the scientific question. The purpose of QSP communication is to explain what the model contributes to a biological or development question.
  • Separate observations from predictions. Model-generated trajectories and simulated outcomes should never be presented as though they were experimental measurements.
  • Choose decision-relevant outputs. A complex QSP model can contain many variables, but only a subset usually needs to appear in the primary communication.
  • Show uncertainty. A central prediction without its relevant uncertainty can create a misleading impression of precision.
  • Explain mechanisms. QSP results are especially valuable when the predicted outcome can be connected to intermediate biological processes.
  • Interpret sensitivity in context. An influential parameter for one output and parameter range is not automatically the most biologically important parameter in every context.
  • Identify extrapolation. Predictions outside the range of supporting observations should be clearly identified as more dependent on model assumptions.
  • Use precise language. Prefer "the model predicts" or "the model supports a mechanistic interpretation" when describing model-generated evidence.
  • Make the analysis reproducible. Model version, parameters, scenarios, simulation settings, and output definitions should be documented even when they are not shown on the main slide.
  • End with the scientific implication. The strongest QSP communication connects the model result back to the original question while clearly stating what remains uncertain.
Next step

Where to Go Next

A natural next step is to study QSP model qualification and credibility, including how model purpose, context of use, verification, validation, sensitivity analysis, uncertainty analysis, and external evidence support confidence in model predictions.

From there, related topics include model qualification for QSP, credibility assessment for mechanistic models, Bayesian calibration, parameter estimation, structural and practical identifiability, and sensitivity analysis.

← Back to Pharmacokinetics Tutorials