Skip to content

Lab: NLME and PKPD Modelling

This is the third computer exercise in the course TBMT42 - Systems biology, Digital Twins and AI.

In this computer exercise you are a drug discovery scientist asked to apply non-linear mixed effect (NLME) modelling and allometric based interspecies translations to assess the mysterious substrate X (SubX1). The assessment will be based on a limited amount of data and may force you to make several assumptions. It is important that you communicate these assumptions and your findings in a lab report, as it will be the foundation for whether or not your company should continue to invest resources into the substrate.

In practice this will be done in three steps:

  • Step 1: Investigate pharmacokinetics and pharmacodynamics (PKPD) of a drug from two rat studies.
  • Step 2: Use the insights from step 1 to find scaling constants from rat to non-human primate (NHP) data.
  • Step 3: Use steps 1-2 to make human drug treatment predictions.

Everyone in your team is waiting for your analysis - Should we continue investing resources into the substrate? It's all up to you.

Throughout this exercise you will encounter different admonitions, which is a type of text box. A short legend of the different kinds of admonitions used can be found in the box below:

Different admonitions used (click me)

Background and Introduction

Useful information

Here you are tasked to do something

Reflective questions

This computer exercise is developed in MATLAB and it is recommended to complete the exercise in MATLAB. This is because MATLAB has many pre-made functions which are useful when you do NLME modelling. Please note that the practical instructions below will focus on an implementation in MATLAB and the supplied code is written in MATLAB. However, the method can be implemented in any environment of your choosing. If you choose to proceed with the computer exercise in another environment/software than MATLAB you will have to compile the tools and re-write the pre-made scripts yourself.

This computer exercises contains solutions and toolboxes that requires the following to be installed before you start:

  • MATLAB running on your computer (or a computer in the computer hall)
  • A C-compiler installed in MATLAB
  • [IQM tools] downloaded and installed, see instructions, in MATLAB. This tool is included in the folder where you download the supplied data and code.

In MATLAB you'll want the following toolboxes installed:

  • Global optimization Toolbox
  • Statistics and Machine Learning Toolbox

To complete this exercise you are supplied with both code and data, download: Lab 3 files. Download these files and insert them into your MATLAB directory. When done please continue to read.

How to pass this computer exercise

To pass this computer exercise you need to follow the instructions herein, write a report, and upload the report to the TBMT42 course page on Lisam.

The report should start with a short abstract (max 300 words) where you describe what you have done and what your recommendation is for the future of the substrate that you are analysing, e.g. continue or stop testing the drug. The criteria for your decision are declared below (see "Email with instructions for SubX1 analysis"). After the short abstract you should answer all the questions listed at the bottom of this page. Note that throughout this exercise there are more questions listed in the task boxes. These questions are there to help you understand the exercise, but if they aren't listed at the bottom of this page then you do not need to provide the answers to them. The questions at the bottom of the page should be answered in a list format where you give both question and your answer to it. In your report, please include relevant figures that you have produced throughout this exercise.

Introduction

You are a drug discovery researcher employed in a team which is trying to produce a drug that can decrease HormoneX. However, all drug candidates up to this point have been rejected. This is because the candidates do not pass your teams criteria for what makes a drug useful, safe, and affordable.

Today you got an email sent from your teams’ project leader that reads:

Email with instructions for SubX1 analysis

Hi,

We have new data on our latest HormoneX repressing drug candidate. We call this new candidate SubX1. Can you do your mathematical-modelling-magic to see how a drug treatment might look in a human from our animal studies?

Currently, we have a limited amount of data on SubX1, but we have performed three smaller studies:

  • Study 1: IV 80 umol/kg administrations in 10 rats. Data is on both SubX1 plasma concentrations and HormoneX levels. Their weights (kg) from rat 1 to 10 are: [0.3,0.28,0.29,0.32,0.33,0.33,0.31,0.27,0.31,0.34].
  • Study 2: Oral 80 umol/kg administration in 10 rats. Data is on SubX1 plasma concentrations. The rat subjects are the same as in study 1 and declared in the same order e.g. rat 1 in study 1 is also rat 1 in study 2 etc.
  • Study 3: IV 80 umol/kg administration in one NHP. Data is on both SubX1 plasma concentrations and HormoneX levels. The bodyweight of the NHP is 4.31 kg.

All this data, our previous HormoneX decreasing mathematical model, and its corresponding MATLAB scripts are supplied here: Lab 3 files

If we are going to proceed with SubX1, we need an oral administrated human treatment that fulfils the following criteria:

  1. The toxicology team concluded that SubX1 plasma concentrations can never be higher than 130 umol/L. Above 130 umol/L negative side effects start to occur.

  2. The biology and metabolism team concluded that HormoneX must be reduced by 50% from the patient's basal levels for the health benefits to take effect.

  3. The production team concluded that we cannot move forward if the treatment requires more then 0,2 mol/week/patient.

  4. The marketing team concluded that we cannot have a pill which needs to be taken more than twice a day.

We have a meeting with the board in a week, please have your analysis ready for then.

Good luck,

project leader

In the downloadable folder referred to in the email you will find:

  • Data, located in the folder Data.

  • A mathematical model called SubX1_model.txt, located in the main folder. This mathematical model was created by a colleague during the development of a previous HormoneX decreasing candidate.

  • The toolbox used to simulate the SubX1_model.txt, located in the folder IQMtools. This toolbox should already be installed on your computer. If not, please read the installation.txt found in the IQMtools folder.

  • MATLAB code, which you can be used as a foundation for your analysis. This code was created by a colleague during the development of a previous HormoneX decreasing candidate.

There are many different ways you can solve your task. This exercise will exemplify the following solution:

  • Step 1: Investigate the two rat studies (study 1 and 2). Here, you will identify pharmacokinetic (PK) and pharmacodynamic (PD) properties of SubX1. You will also identify variability between the 10 rats in the studies.
  • Step 2: Find metabolic and volume scaling of the properties identified in step 1. This scaling is done using an allometric scaling to species body weight. The allometric scaling is found using both the rat studies 1-2 combined with a NHP study (study 3).
  • Step 3: Make human drug treatment prediction from step 1-2. Here, you will use everything you have learned to evaluate if SubX1 may pass all the criteria your project leader specified in the email above.

In this three-step method there are many simplifications and assumptions which are interesting to discuss, explore, and/or improve. Simplifications and assumptions are a necessary part of the analysis and this exercise. You are encouraged to change these assumptions to your own hypothesis - if something doesn't work then you can always return to the original pre-written method.

Best of luck and have fun!

Getting familiar with the data and model

Open the main.m script. The main script runs all other scripts to ensure that they are done in the correct order. When running a script from the main it is important to remove the # symbol. When you do not want to run a script, it is important to add the # symbol. The first script that the main.m calls for is LoadThings.m, which is a script designed to always run when the main.m is used.

Open the LoadThings.m script. Get familiar with what the script does. Everything you write here is intended to run every time you call on something through the main.m script. Typical things that should be included here are setting up your workspace, setting up the model you work with, and declaring variables that all scripts require, e.g. colors, index names, species body weight, etc.

To get familiar with the data you are supplied with the script PlotStudies.m. Open this script and get familiar with both the code and the data (see Task 1).

Task 1: Data processing

Task 1A: Plot the pharmacokinetics of SubX1

Run the PlotStudies.m script through the main and identify the pharmacokinetic data plot of the three studies. One typical pharmacokinetic plot is to have drug concentration on the y-axis and time on the x-axis. Try to get a feeling of the pharmacokinetics from these plots, for example: Why is the maximum exposure of SubX1 much higher in study 1 than study 2? Why is the exposure at the end higher in study 2 compared to study 1? Why is the exposure of SubX1 so similar in study 3 compared to study 1, even though the IV-dose in study 3 is about 344.8 umol and the IV-dose in study 1 is 24 umol?

What is pharmacokinetics?

Pharmacokinetics includes drug rate of appearance, rate of disappearance, and distribution. It can be described/simplified as "what the system does to the drug".

Task 1B: Plot the pharmacodynamics of SubX1

Identify the pharmacodynamic data plot of the three studies. One typical pharmacodynamic data plot is to have the effect of the drug on the y-axis and time on the x-axis. Try to get a feeling of the pharmacodynamics from these plots, for example: how fast or slow do HormoneX decrease after drug administration? How quickly does it return to baseline? If you were to give a higher dose of SubX1 do you think it would be able to remove more HormoneX?

What is pharmacodynamics?

Pharmacodynamics includes the effect of the drug. It can be described/simplified to "what the drug does to the system", which in this case is how HormoneX changes.

Task 1C: Plot the pharmacokinetic and pharmacodynamic (PKPD) of SubX1

Identify the PKPD data plots. One typical PKPD plot has pharmacodynamics on the y-axis and pharmacokinetics on the x-axis. Try to understand the PKPD curve: What are those outliers in the top right corner which seems to not follow the same trend as the others, and why are they there? Can you visually estimate the drug EC50 (drug exposure at half effect), and Emax (maximum effect).

After getting familiar with the data, it is time to get familiar with the mathematical model. Open SubX1_model.txt file in the main folder and try to understand the mathematical model. Below is a description of all parts of the model:

States:

  • State "A" is an administration state which represents the delay between a drug SubX1 administration (umol) and when it is observed in the body. This delay could describe many things, e.g. a subcutaneous (SC) injection, but in this example represents the delay from oral administrations.
  • State "C1" stands for "Central state 1", which represents drug concentration (umol/L) in the plasma.
  • State "HormoneX" represents relative hormone levels to subject baseline (%).
  • State "mod" is a timer which can be used to declare repeating drugs administrations. This is useful if you want to simulate drug administrations over many days. These repeated administrations are simulated because of a loop declared in MODEL EVENTS. This loop resets “mod” and administer a new drug dose when “mod” reaches the value of "ReAdministration_Time". The drug dose in this loop is declared with "ReAdministration_Amount". Make sure to keep the value "ReAdministration_Time" higher than the length of your simulation if you do not want re-administrations to occur.

Parameters:

  • Ka - rate of absorption (1/min)
  • V1ref - volume of central compartment C1 (L)
  • CLref - rate of drug loss (1/min)

  • f - bioavailability (%)

  • Emax - maximum drug effect (fold-increaase from baseline)

  • EC50 - drug potency (umol/L)

  • Rb - metabolic scaling constant, scaling reaction rates to body weight (dimensionless)

  • Vb - metabolic scaling constant, scaling volumes to body weight (dimensionless)

  • BW - total body weight (kg)

  • ReAdministration_Time - time between re-administrations (min)

  • ReAdministration_Amount - drug dose of re-administrations (umol)
How do SubX1 affect HormoneX in the model?

In the model SubX1 affects HormoneX by increasing the negative derivative term of the state "HormoneX". The increasing negative term represents a sum of all pathways a hormone can decrease, e.g. active clearance, excretion, degradation etc. The increased negative term is done through the MODEL REACTION "SubX1_PD".

The parameter values listed in the SubX1_model.txt model is used as a foundation to describe all species simulated in the model. Some reactions and all volumes are expected to be different between subjects and species. These differences can be described with scaling of fixed effects and random effects.

Fundamentals of interspecies translations: Allometric scaling

A fixed effect describes all species and participants in the model, and a scaling of a fixed effect can be based on a co-variable that alters the fixed effect based on e.g. body weight. If the covariable is an allometric variable, then the process is called allometric scaling. A commonly used co-variable for the type of analysis you do in this task is an allometric scaling based on total body weight. The allometric scaling in the SubX1_model.txt model is done in the MODEL VARIABLES section.

Fundamentals of subject-specific translations: Random effects

A random effect changes a parameter value to describe a specific subject. This change is typically only data-driven, which means it is not based on co-variables but only determined by data. One of the main features of NLME modelling is to create and find these random effects. In NLME random effects can be restricted to being normally distributed around the fixed effect. This restriction is often useful when you are describing something that you expect to be normally distributed in real-life. Random effects cannot be seen in the model .txt file, as they will be created when you train the model to the data.

Allometric scaling is standard practice in drug development research and clinical applications. An example of clinical application of allometric scaling in your everyday life is when your doctor prescribes a drug treatment dependent on your total body weight, this is typically done with a 1:1 ratio (dose/kg). In drug development research the allometric scaling used is typically more complex: i) we do not expect both volumes and metabolic rates to scale equally to body weight, and ii) we do not expect both volumes and metabolic rates to scale with a ratio of 1:1. Instead, the scaling is expected to follow a logarithmic relation which is often called Kleiber's law.

What is Kleiber's law?

According to Kleiber's law, the metabolic rates is expected to increase around ParameterRef(BWact/BWref)^0.75, and for volumes around ParameterRef(BWact/BWref)^1. Where "ParameterRef" is the reference parameter value (which is the parameter you want to scale to body weight), where "BWref" is the reference body weight (animal with the lowest body weight; in this example rat ), where "BWact" is the body weight of the subject you are scaling the volume or metabolic rate to (in this example it may be a specific rat, NHP, or human), and finally the value of 0.75 or 1 (which is a constant which can be referred to as the Kleiber's constant; note that 0.75 is used for metabolic rates and 1 is used for volumes).

Now, you should be familiar with the data and the mathematical model. It is time to combine them to analyze study 1 and 2, see step 1.

Step 1

Open PlotRatExperiment_InitialGuess. This script will help you to simulate study 1 and 2 with the parameters declared in the SubX1_model.txt model file. Note that these parameter values describe a previous HormoneX reducing candidate and will not describe the new SubX1 candidate well. Read the script and when familiar with the code, proceed to Task 2.

Task 2: Simulate and compare initial guess to study 1 and 2

Task 2A: Simulate and plot your simulations of the rat studies 1-2

Simulate study 1 and study 2. To aid you with this task use the script PlotRatExperiment_InitialGuess. Note that Study 1 is administrated by IV and study 2 is administrated orally. Also note that all rats have different body weight. Compare your initial simulation to the data.

Task 2b: Manually calibrate the parameter values to the data

Try to understand why the simulation does not fit the data and change parameter values in the model file to see if you can make it fit better. Take your time to guess how the simulations will change when you change a parameter value e.g. how will the pharmacokinetics change if you increase CLref 2-fold? Note that you are not expected to find perfect fit to data this way, but the exercise is rather to get a feeling of how parameter values change the model dynamics. Try to imagine what parameters could need random effects (variability in parameter values) to describe the variability between the rats (variability in data).

Task 2C: Save the plots from Task 2B

Save the pharmacokinetic and pharmacodynamic plots produced with your manually calibrated parameters (Task 2b).

As seen in the rat studies, there are fundamental and natural variability between the different rat-subjects which are not fully described by their differences in total body weight. Therefore, we will now use NLME modelling to create random effects which can describe these differences, see task 3.

Task 3: Optimise the model parameters to describe studies 1-2 and plot your agreement to data.

Task 3A: Run a NLME optimisation to describe studies 1-2.

To succeed with this task, you have been supplied with the script OptimiseRatExperiment which is found in the Step1_PopulationVariability folder. Open the script and get familiar with the code. You are welcome to change the optimization settings and NLME optimization method if you want to e.g. you can improve the method by removing variability in parameters which do not need it or adding variability in new parameters. When the NLME optimization finishes you will be provided with the outputs PHI, PSI, b, and PHIR which will be saved as a file named OptimisedParams_Step1. The variables PHI, PSI, b, and PHIR meaning can be derived from the code combined or from the nlmefit documentation. If you need more aid, please see function documentation on the MATLAB webpage or by writing help nlmefit in the terminal. You can also look at the script PlotRatExperiment in the Step1_PopulationVariability folder to see how the different variables are used after the optimization.

Note that your manual calibration is the starting guess for your optimisation.

Where is PHI, PSI, b, and PHIR created?

When the NLME optimization script is finished then you will see PHI, PSI, b, and PHIR in your workspace. PHI, PSI, and b are outputs from the NLME optimization and PHIR is calculated after the NLME optimization from PHI and b.

Short description on how the OptimiseRatExperiment function works

The OptimiseRatExperiment calls for a NLME optimization. The NLME optimization loops over the SimulateRatExperiment function to simulate all the subjects and minimize the differences between the variable Y (found in the optimization script) and the variable varargout (found in the simulation script). In the OptimiseRatExperiment the variable GROUP is declared, which is a vector that supplies the NLME optimization with information of which datapoints are from which subject. In the shared NLME optimization all subjects are supplied with unique random effects for the parameters which is allowed to have random effects. The vector REParamsSelect determines which parameters are allowed to have a random effect. All parameters not declared in the REParamsSelect vector will be restricted to only having one fixed effect value to describe the full population. The global variable GroupID is one solution to add allometric scaling to the NLME optimization, where each rat is simulated with different total body weight (parameter BW). Please feel free to give more or less parameters variability and remember to add the variability to the fixed effects after the optimization, see the calculation of PHIR to do so.

Understanding the NLME optimisation output
  • What does PHI, PSI, b, and PHIR represent? What are they used for?
Task 3b: Plot the model agreement to data, parameter variability, and a PKPD curve after the NLME optimisation

To succeed with your task, you have been supplied with the script PlotRatExperiment which is found in the Step1_PopulationVariability folder. Take a minute to understand the code. After you have read the code, compare the pharmacokinetics to the pharmacodynamics. Do the rats display different PKPD curves? If so, how do you capture these differences in the model? Please find what parameters have random effects and comment on the variability of the parameters, e.g. best/max/min value and/or variability/mean. Is the identified variability high? Could a high variability be a problem for your future human drug predictions?

With a model describing studies 1-2 you have successfully found parameter variability and overall pharmacokinetic and pharmacodynamic properties. You may now proceed to step 2.

Step 2

You can now use the result from Step 1 to make predictions of species with other bodyweights. However, the allometric scaling will not be optimal without a calibration to a population with a larger variability in body weight. This is because in our example rat-rat variability will create too much noise to accurately estimate how their differences in body weight affected the PKPD. Before this calibration you can assume that the metabolic rates and volumes follow "Kleiber's law". You will now make a prediction of the final study you have - study 3 - which is an IV administration in a non-human primate (NHP).

Now, it is time to use your results from step 1 to predict study 3 - the NHP study, see task 4.

Task 4: Predict study 3

To succeed with your task, you have been supplied with the script PlotMetabolicScaling_InitialGuess which is found in the Step2_MetabolicTranslation folder. Go through the script and try to understand as much as you can. See if there are any assumptions you want to change in the prediction e.g. metabolic rate scaling (Rb), volume scaling (Vb) etc. Most importantly, you will find one plot of the allometric scaling with parameter values on the y-axis and total body weight on the x-axis (kg): Where is the slope of the scaling derived from? Does the scaling formulate an area of a line and why do we need both?

Compare your predictions to the NHP study to the data.

Now it is time to make one final NLME optimization. This time the NLME will be done to all the data you have available (studies 1-3), see task 5.

Task 5: Run a new parameter estimation to describe studies 1-3 and plot the new description

Task 5A: Run a NLME optimisation to describe studies 1-3

To succeed with this task, you have been supplied with the script OptimiseMetabolicScaling which is found in the Step2_MetabolicTranslation folder. Open the script and get familiar with the code. You are welcome to change the optimization settings and NLME optimization method if you want to. Remember that if you made any changes to the simulation assumptions in task 4 then you need to do them also in the script used in the NLME optimization to simulate the studies SimulateMetabolicScaling.

Task 5B: Plot and save your results from the NLME optimisation.

To succeed with your task, you have been supplied with the script PlotMetabolicScaling which is found in the Step2_MetabolicTranslation folder. Compare the results from this optimization and the optimization in Step 1 e.g. difference in scaling constants, difference in parameter variability etc. Did the description of the rat PKPD properties change when you added the NHP data? Most importantly, what is the new value of "Rb" and "Vb"? Is the value of "Rb" and "Vb" what you expected?

You have now used the rat data to quantify natural population-based variability and PKPD properties (Step 1), and you have found metabolic rate and volume correlation to total species body weight (Step 2). You are now ready for the final step in this computer exercise which is making a human drug prediction without human drug data (Step 3).

Step 3

You are now close to the end of your analysis, and your project team is waiting anxiously for your conclusion. In the folder Step3_ClinicalTesting you will find the script PredictClinicalTesting which will help you in making the human drug treatment predictions. These predictions should be made over several days to capture longer-term dynamics e.g. drug accumulation. In the supplied code the length of the simulation is 2 weeks, but feel free to increase or decrease it as you like.

Note that one of many fundamental differences between species is bioavailability. Bioavailability has no found correlation to total body weight, and therefore it is not obvious how you should take bioavailability into consideration in your analysis. You do however need to assign a value and motivate what value you choose and why when making your human predictions.

To aid you in this decision please mind the following:

  • In your description of study 1-2, you have estimated the bioavailability for oral administrations in rats.
  • A previous HormoneX reducing drug candidate with a similar molecular structure to SubX1 displayed an orally administrated bioavailability of 62 % in NHP.
What is bioavailability?

Bioavailability can be simplified and reduced to a percentage of how much of a drug that enters the body reaches the plasma.

Historically, just like bioavailability, drug potency (EC50) has not displayed any significant correlation to total body weight. To overcome this limitation, modern drug development typically uses in vitro studies to estimate drug potency. In your case you do not have such in vitro measurements. Therefore, you are free to assume drug potency from study 1-3 however you think is appropriate.

Look over the PredictClinicalTesting script, try to understand what it does, and make any changes to the prediction method as you like. There are especially four variables which are crucial for you to look at:

  • HumanBW: which is the assumed body weight (kg) of the human-participant you are simulating.
  • BioavailabilityHuman: which is assumed oral bioavailability in humans (%).
  • ReAdministration_Time: Which is how often (min) the human is consuming SubX1 orally.
  • ReAdministration_Amount: Which is the dosage of active substrate (SubX1) in the drug (umol).

Change these variables and investigate how things change in your predictions. Think about more assumptions that you make when doing these human simulations. When you feel ready, it is time to see if SubX1 fulfils the human treatment drug requirements specified in the email from your project team leader (see emailed instructions in the introduction section), see task 6.

Task 6: Test different treatments and save an acceptable treatment or the best treatment you can find.

Save a treatment that fulfils all the criteria listed in the instructions (see introduction section above). If your prediction of SubX1 in human cannot fulfil the criteria then save the treatment that you can produce that is the closest to the criteria. Note that your analysis quality is measured in the plausibility of your assumptions and the motivation you make regarding your predictions - not whether the candidate is rejected or not.

Reflective questions
  • Do you believe SubX1 is good enough to invest more resources into?
  • What pre-clinical data do you feel would be the most valuable for you to make better predictions?
  • Were you surprised by the effect of changing some variables e.g. increasing dosing frequency or giving the same dose to people with different body weights?
  • What drug potency (EC50) do you use to make the human prediction, and what is it derived from?
  • What drug bioavailability (f) do you use to make the human prediction, and what is it derived from?
  • What drug clearance (CL) do you use to make the human prediction, and what is it derived from?
  • What drug plasma volume (V1) do you use to make the human prediction, and what is it derived from?

You are now done with the modelling work! Well done!

Your final task is to write your analysis before the big board meeting (a.k.a computer exercise report).

What should your computer exercise report include?

Your report should start with a short abstract (max 300 words) which summarizes what you have done and if you think your team should continue to invest money and time into SubX.

After the abstract you need to answer the following questions below. Make sure to write both the question and answer. Remember to include relevant figures that you have saved throughout this computer exercise.

Task 1 - Data processing

  • Question 1.1: What information can a PKPD curve give?
  • Question 1.2: Why does the SubX exposure in Study 2 not reach the same maximum concentrations as study 1?
  • Question 1.3: What are those outliers in the top right corner of the PKPD plot, which seems to not follow the same trend as the others? Why are they there?

Task 2 - Simulate and compare initial guess to study 1 and 2

  • Question 2.1: If your task was only to describe the rat data (studies 1-2), what would have been the best tool: allometric scaling based on total body weight, or random effects?
  • Question 2.2: Give an example of a co-variable you think could have been useful when scaling our metabolic rates to rats. Because we have used body weight in this example you are not allowed to use it.
  • Question 2.3: Show your simulation of study 1 and 2 with your manual calibration i.e. the simulation used as your starting guess for the parameter optimization.

Task 3 - Optimise the model parameters to describe studies 1-2 and plot your agreement to data

  • Question 3.1: What do the PHI, PSI, b, and PHIR variables represent from the NLME optimisation?
  • Question 3.2: Show the model agreement to data after the optimisation.
  • Question 3.3: Did all the parameters you tested need a random effect to be able to describe the data? Did any parameter display a larger need of variability? Show the parameter values used to describe all rats.

Task 4 - Predict study 3

  • Question 4.1: What is Kleiber's law, and how do you use it to predict study 3?
  • Question 4.2: Show your initial guess of study 3.

Task 5 - Run a new parameter estimation to describe studies 1-3 and plot the new description

  • Question 5.1: What values did your allometric scaling variables (Rb and Vb) get after the optimization? If you don’t have enough information to identify the Rb and Vb values, what is a typical initial guess you can use? What are Rb and Vb used for?
  • Question 5.2: In your work, which parameters were given random effects, which parameters were scaled with allometric scaling, and which were described using both?
  • Question 5.3: Show the model agreement to data after the Step 2 optimization.

Task 6 - Test different treatments and save an acceptable treatment or the best treatment you can find

  • Question 6.1: What happens if you simulate the same amount of SubX1 every day, but with smaller and more frequenct administrations?
  • Question 6.2: What drug potency (EC50) do you use to make the human prediction, and how do you motivate it?
  • Question 6.3: What drug bioavailability (f) do you use to make the human prediction, and how do you motivate it?
  • Question 6.4: What drug clearance (CL) do you use to make the human prediction, and how do you motivate it?
  • Question 6.5: What drug plasma volume (V1) do you use to make the human prediction, and how do you motivate it?
  • Question 6.6: What pre-clinical data do you feel would be the most valuable for you to make better predictions?
  • Question 6.7: What is your final decision on the future of SubX1? Describe the treatment schedule you produced that either fulfilled the criteria (see introduction) or was the closest to fulfilling the criteria.