Home

User Manual - Laboratory of Applied Pharmacokinetics

Contents

1. T ag1 ag type MA MA MA Vt MA MA MA Vt MA MA MA Vt MA MA MA PMmatrix summary x summary PMfinal VtMed vVtMed vVtMed VD VD VD Med VtMed VtMed vVtMed Med vVtMed Med vVtMed VtMed VD YD VD 0 U Q Qe 025 5 s979 0235 i3 2975 2025 ao 875 4025 A uc 025 NS 875 0253 E 979 formula quantile FUN valu OSE 48E 19 cA pE 07E ead 39 481 94 48 18 47E 29E E 03 e E 02 GI 00 229 07E 2738 88E ISE ET02 E 00 TOE ba 7 02E pM In this example the weighted median for Tlag1 is 1 11 with a 9596 CI around the weighted median of 0 0688 to 1 79 The median absolute weighted difference MAWD is 0 502 with a 9596 CI of 0 17 to 0 789 oF E 01 include exclude Pmetrics User s Guide 29 This function will summarize a Pmetrics data file which is of class PMmatrix when loaded by PMreadMatrix or PMload The simplest is to summarize just the object For example data PMex1 summary mdata 1 umber of subjects 20 umber of inputs 1 umber of outputs 1 otal number of observations outeq 1 139 with O 0 000 missing umber of covariates 5 ZA
2. e qgrowth This function will return a dataframe of height and weight for boys girls or both for a given range of ages in months and body size percentile e g the median This can be useful for simulations in Pmetrics The data for this function come from the 1977 National Center for Health Statistics Growth Chart Equations on the website of the United States Centers for Disease Control e SS PK Sample size calculations for PK studies are difficult as traditional power analysis is based on comparisons between 2 or more groups One proposed method is to sample sufficient subjects to ensure that the width or precision of the standard error of the mean value for a given PK parameter is within a specified degree for a given probability This function uses the formula for precision of the standard error of the mean of a distribution of PK parameter values to calculate either n the sample size or sd the maximum standard deviation of a PK parameter value distribution for a given mean desired degree of precision and confidence The formula is n Z 1 ciy2 sd precision mean where Z is the standard normal quantile ci is the probability of the confidence interval e g 0 95 sd is the standard deviation of the parameter values precision is the desired width of the confidence interval as a proportion of the mean e g 0 2 2096 and mean is the mean parameter value References 1 Goutelle S Bourguignon L Maire PH Van Guilder M Conte JE
3. as can conditional statemtents in Fortran code A prefix is not necessary in this block for any statement although if present will be ignored Example F FA 1 FA1 Lag time Specify the lag term if present which is the delay after an absorbed dose before observed concentrations Use the form TLAG expression where is the input number Primary and secondary variables and covariates may be Pmetrics User s Guide 12 eC used in the expression as can conditional statemtents in Fortran code A prefix is not necessary in this block for any statement although if present will be ignored Example Lag TLAG 1 Tlag1 Differential equations Specify a model in terms of ordinary differential equations in Fortran format XP is the notation for dX dt where is the compartment number X is the amount in the compartment There can be a maximum of 20 such equations Example Dif XP 1 KA X 1 XP 2 RATEIV 1 KA X 1 KE KCP X 2 KPC X 3 XP 3 KCP X 2 KPC X 3 RATEIV 1 is the notation to indicate an infusion of input 1 typically drug 1 The duration of the infusion and total dose is defined in the data csv file Up to 7 inputs are currently allowed These can be used in the model file as RATEIV 1 RATEIV 2 etc The compartments for receiving the inputs of oral bolus doses are defined in the Bolus block Outputs woe Output equations in Fortran format Outp
4. compartments times are reset to 0 as for an EVID 4 event Il This is the interdose interval and is only relevant if ADDL is not equal to 0 in which case it cannot be missing If ADDL 0 or is missing II is ignored INPUT This defines which input i e drug the DOSE corresponds to Inputs are defined in the model file OUT This is the observation or output value If EVID 0 there must be an entry if missing this must be coded as 99 It will be ignored for any other EVID and therefore can be There can be at most 150 observations for a given subject OUTEQ This is the output equation number that corresponds to the OUT value Output equations are defined in the model file CO C1 C2 C3 These are the coefficients for the assay error polynomial for that observation Each subject may have up to one set of coefficients per output equation If more than one set is detected for a given subject and output equation the last set will be used If there are no available coefficients these cells may be left blank or filled with as a placeholder COV Any column after the assay error coefficients is assumed to be a covariate one column per covariate Manipulation of CSV files There are several functions in Pmetrics which are useful for either converting other formats into Pmetrics data files or checking Pmetrics data files for errors and fixing some of them automatically PMreadMatrix filename This function simply reads
5. standard deviation and covariance matrix of the distribution and perform a standard Monte Carlo simulation The second way is what we call semi parametric and was devised by Goutelle et al 1 In this method the non parametric support points in the population model each a vector of one value for each parameter in the model and the associated probability of that set of parameter values serve as the mean of one multi variate normal distribution in a multi modal multi variate joint distribution The weight of each multi variate distribution is equal to the probability of the point The overall population covariance matrix is divided by the number of support points and applied to each distribution for sampling Limits may be specified for truncated parameter ranges to avoid extreme or inappropriately negative values When you load simulator output with STMparse it will include values for the total number of simulated profiles needed to generate nsim profiles within the specified limits as well as the means and standard deviations of the simulated parameters to check for simulator accuracy Output from the simulator will be controlled by further arguments to SIMrun If makecsv is not missing a csv file with the simulated profiles will be created with the name as specified by makecsv otherwise there will be no csv file created If outname is not missing the simulated values and parameters will be saved in a txt file whose name is that s
6. 1 plot type vpc An example of this plot is shown below Visual Predictive Check out time External Validation Should you wish to use your population model to test how well it predicts a second population that is separate from that used to build the model i e externally validate your model you may do that in Pmetrics After completing an NPAG run place the same model file located in the inputs subdirectory of the NPAG run whose model you are validating along with your new validating data file in your working directory So there should be two files in your working directory e model txtfile This will be the same as for model building NPAG run found in the inputs subdirectory e data csv file This will be a Pmetrics data input file containing the new subjects for validation Next do the following steps to complete the validation NPAG run 1 Load the model building run with PMload run numl so that its NPdata object is in memory 2 Initiate an NPAG run in Pmetrics as usual but with an additional argument to specify the model density file which will serve as a non uniform prior e g NPrun model mymodel txt data validation csv prior NPdata 1 cycles 0 where NPdata 1 is an example of the object loaded in step 1 in this case with PMload 1 Specifying 0 cycles will calculate a Bayesian posterior only for each subject in the validation data set Complete the NPAG run as usual 4 Load the result
7. CI These estimates correspond to weighted mean 9596 CI of the mean variance and 9596 CI of the variance respectively for a sample from a normal distribution To estimate these non parametric summaries the function uses a Monte Carlo simulation approach creating 1000 x npoint samples with replacement from the weighted marginal distribution of each parameter where npoint is the number of support points in the model As an example if there are 100 support points npoint 100 and for Ka there will be 1000 sets of 100 samples drawn from the weighted marginal distribution of the values for Ka For each of the 1 000 sets of npoint values the median and MAWD are calculated with MAWD equal to the median absolute difference between each point and the median of that set The output is npoint estimates of the weighted median and npoint estimates of the MAWD for each parameter from which the median 2 5th and 97 5th percentiles can be found as point estimates and 9596 confidence interval limits respectively of both the weighted median and MAWD For IT2B runs the function will return the mean and variance of each parameter and the standard errors of these terms using SE mean SD sqrt nsub and SE var var sqrt 2 nsub 1 Pmetrics User s Guide 28 eC Example data PMex1 summary final 1 pa Ka Ka Ka Ka Ka Ka Ke Ke Ke Ke Ke lt lt oW d d d E agi lagl lagl lagl T
8. If you use a for file directly it will be in the inputs subfolder of the run folder not in etc since you did not use the simpler template as your model file Structure of model files The new model file is a text file with 11 blocks each marked by followed by a header tag PRImary variables COVariates SECcondary variables BOLus inputs INItial conditions F bioavailability LAG time DIFferential equations OUTputs ERRor EXTra For each header only the capital letters are required for recognition by Pmetrics The blocks can be in any order and header names are case insensitive i e the capitalization here is just to show which letters are required Fortran is also case insensitive so in variable names and expressions case is ignored Details of each block are next followed by a complete example Important Sometimes it is important to preserve spacing and formatting in Fortran code that you might insert into blocks particularly the EXTRA block If you wish to do this insert format and format before and after any code that you wish to reproduce verbatim with spacing in the fortran model file Comments You can insert comments into your model text file by starting a line with a capital C followed by a space These lines will be removed ignored in the final fortran code Primary variables Primary variables are the model parameters that are to be estimated by Pmetrics or are design
9. Jelliffe RW Population modeling and Monte Carlo simulation study of the pharmacokinetics and antituberculosis pharmacodynamics of rifampin in lungs Antimicrob Agents Chemother 2009 53 2974 2981 Pmetrics User s Guide 53 eC 2 D Agostino R Transformation to Normality of the Null Distribution of G 1 Biometrika 1970 57 679 681 Mentr F Escolano S Prediction discrepancies for the evaluation of nonlinear mixed effects models J Pharmacokinet Pharmacodyn 2006 33 345 367 Pmetrics User s Guide 54
10. a link to download and install Apple s Xcode application if you do not already have it on your system Xcode is required to run gfortran on Macs As of version 4 3 for Lion Xcode is available from the App store for free For Snow Leopard Xcode is on your installation disk NOTE For Xcode downloaded from the App store Lion and later you must additionally install the Command Line Tools available in the Xcode Preferences gt Downloads pane Lion and Mountain Lion or the Apple Developer website for Xcode 5 Mavericks Windows users need to pay special attention because the the gcc installer that provides necessary common libraries for many programming languages does not by default include gfortran When gcc is installed be sure to choose the fortran option to include gfortran as shown below TOM GCC Setup E ete New Installation Ch Choose which fe TDM GCC Setup EC EI New Installation Choose Components Choose which features of TDM GCC you want installed Check the components you want installed and uncheck the components you don t want installed Click Instal to start the installation Check the components you want installed and uncheck the components you don t want installed Click Install to start the installation Select the type of install TDM GCC Recommended C C X Select the type of install Custom jCheck this box Or select the optional components y
11. ainan TEE MM FERE M ERN EE OR ER UE FEFERM arsa ESRO DUMP UE 49 Internal EIETILME 49 External Validation 1 1 1 eei iii otto etes it Saee urne p baba E EME SEN NR a EEE NEEESE ee 52 dulsipeienclmeU I 52 Ancillary Furictlons os cocien eio cie bene anl ne ban xd kin d Du lh dedil ea v pud gw E n UU DO 53 IZ NETTE TT TETTE OTT 53 Pmetrics User s Guide 2 Introduction Thank you for your interest in Pmetrics This guide provides instructions and examples to assist users of the Pmetrics R package by the Laboratory of Applied Pharmacokinetics at the University of Southern California Please see our website at http www lapk org for more information Here are some tips for using this guide Items that are hyperlinked can be selected to jump rapidly to relevant sections At the bottom of every page the text User s Guide can be selected to jump immediately to the table of contents Items in courier font correspond to R commands Citing Pmetrics Please help us maintain our funding to provide Pmetrics as a free research tool to the pharmacometric community If you use Pmetrics in a publication you can cite it as below In R you can always type citation Pmetrics to get this same reference Neely MN van Guilder MG Yamada WM Schumitzky A Jelliffe RW Accurate Detection of Outliers and Subpopulations With Pm
12. and imprecision Bias is the mean weighted error of predicted observed Imprecision is the bias adjusted mean weighted squared error of predicted observed We adjust for bias because the mean squared error MSE calculated as pred obs for all predictions observations is equal to the sum of the variance of the predictions and the bias of the predictions i e Var pred Bias pred Therefore the true variance or imprecision of the predictions is MSE Bias For both calculations the weighting is according to the calculated SD of the observation using CO C1 C2 and C3 The specific formulae are as follows e Weighted prediction error wpe pred obs SD for each prediction observation e Weighted squared prediction error wspe pred obs SD for each prediction observation e Bias mean weighted prediction error mwpe wpe N e Imprecision bias adjusted mean weighted squared prediction error bamwspe Y wspe N mwpe All these values are returned with summary for PMop objects Both summary and PMcompare willalso provide the root mean squared error RMSE and RMSE RMSE is the square root of pred obs i e an unweighted measure of imprecision used by many but which we favor less given its failure to account for observation weights or adjust for bias The RMSE is the RMSE normalized to the total RMSE over all observations Visual predictive checks and NPDE Two more complex and time consuming options based on
13. by the analytical laboratory The batch files contain all the information necessary to complete a run tidy the output into a date time stamped directory with meaningful subdirectories extract the information generate a report and a saved Rdata file of parsed output which can be quickly and easily loaded into R On Mac Unix systems the batch file will automatically launch in a Terminal window On Windows systems the batch file must be launched manually In both cases the execution of the program to do the actual model parameter estimation is independent of R so that the user is free to use R for other purposes For the Simulator the SIMrun function will execute the program directly within R For all programs the parse functions will extract the primary output from the program into meaningful R data objects For IT2B and NPAG this is done automatically at the end of a successful run and the objects are saved in the output subdirectory as IT2Bout Rdata or NPAGout Rdata respectively For IT2B and NPAG the PMload function can be used to load the either of the above Rdata files after a successful run PMsave is the companion to PMload and can re save modified objects to the Rdata file The PMreport function is automatically run at the end of a successful NPAG and IT2B run and it will generate an HTML page with summaries of the run as well as the Rdata files and other objects The default browser will be automatically laun
14. by the number of covariate entries icen Median default or mean of the covariates and parameter value distributions pop class PMpop Subject identification data frame post class PMpost data frame NPAG only time Time of each prediction at a frequency specified in theNPrun command with a default of 12 minutes icen Median default or mean of the parameter distributions used to calculate the predicted values pred Population prior PMpop or Bayesian posterior eo predictions for each output equation NENNEN block Same as for PMop above NPdata class NPAG list Raw data used to make the above objects Please ITdata class IT2B list use NPparse or ITparse in R for discussion ofthe data contained in these objects mdata class PMmatrix See Pmetrics Input Your original raw data file data frame Files NPDE class PMnpde list Use the command This object contains the information to perform Str NPDE x in graphical and numerical analysis of normalized This object will only be R where x is the prediction distribution errors It is a method of present if you have run run number internal model validation makeNPDE after a run is completed Pmetrics User s Guide 22 sim class PMsim list This object will only be present if you have run makeNPDE after a run is completed parValues Since R is an object oriented language to access the observations in a PMop object
15. copied to a new file with c prepended to the data file name e g data csv gt c data csv Pmetrics User s Guide 39 eC Examples assuming that the covariates wt weight and ccr creatinine clearance are in your data and you are simulating from run 1 SIMrun covariate list cov cov 1 This will simulate all parameters and covariates that were present in run 1 using the covariate means and covariances that existed in the population No limits will be applied SIMrun covariate list cov cov 1 mean list wt 100 Note the list within a list This simulation will be the same as the first example except that the mean weight will now be 100 retaining the same standard deviation as in the original population SIMrun covariate list cov cov 1 mean list wt 100 sd list wt 30 This will now also replace the standard deviation of weight in the population with 30 SIMrun covariate list cov cov 1 limits list wt c 10 90 age c 1 1000 This will use the existing means and covariances of the covariates in the population but will limit simulated values for weight to be a minimum of 10 and maximum of 90 Age will effectively have no limits Other covariates will have limits equal to the limits in the original population SIMrun covariate list cov cov 1 fix c wt age This will simulate with existing means standard deviations and limits for all covariates except weight and age which will be fixed to th
16. feel free to contact us at any time preferably through the Pmetrics forum at http www lapk org or directly by email at contact lapk org This manual is also not intended to be a theoretical treatise on the algorithms used in IT2B or NPAG For that the user is directed to our website at www lapk org Getting Help and Updates There is an active LAPK forum available from our website at http www lapk org with all kinds of useful tips and help with Pmetrics Register separately from your LAPK registration and feel free to post Within R you can also use help command or command in the R console to see detailed help files for any Pmetrics command Many commands have examples included in this documentation and you can execute the examples with example command Note that here quotation marks are unnecessary around command You can also type PMmanual to launch this manual from within Pmetrics as well as a catalogue of all Pmetrics functions Finally PMnews will display the Pmetrics changelog Pmetrics will check for updates automatically every time you load it with 1ibrary Pmetrics Ifan update is available it will provide a brief message to inform you You can then use PMupdate to update Pmetrics from within R without having to visit our website You will be prompted for your LAPK user email address and password When bugs arise in Pmetrics you may see a start up message to inform you of the bug and a patch can be install
17. for example use the following syntax opSpost1Sobs BENE A data frame with id time out outeq columns containing simulated observations at each time and output equation number in the template data file If simulations from multiple template subjects have been combined see Simulator Runs then id will be of the form x y where x is the simulation number and y is the template number A data frame with id time out comp columns containing simulated amounts in each compartment A data frame with id columns containing the parameter value sets for each simulated subject with signifying the columns named according to the names of the random parameters in the model The total number of simulated sets of parameters which may be greater than the requested number if limits were specified in the simulation see Simulator Runs The means of the parameter values in the total simulated sets which can be used as a check of the adequacy of the simulation to reproduce the requested mean values when limits were applied The final truncated set will likely not have the requested mean values The covariances of the parameter values in the total simulated sets which can be used as a check of the adequacy of the simulation to reproduce the requested covariance values when limits were applied The final truncated set will likely not have the requested covariance values Note that you will place an integer corresp
18. for two output equations Each number is a coefficient that allows calculation of the standard deviation of the normal distribution around zero from which the noise is generated such that SD CO C1 sim C2 sim C3 sim where sim is the simulated value This is directly analogous to the assay error polynomial discussed in the section on the model file format If you do not specify obsNoise all coefficients will be set to 0 for all output equations i e no noise If you specify obsNoise NA values in the data file will be used similar to Limits above If these values in the data file are missing values in the model file will be used Examples SIMrun SIMrun obsNoise NA SIMrun obsNoise c 0 0 1 0 0 1 0 15 0 0 The first example sets all the observation noise terms to 0 The second example uses the error terms in the model file The third example sets the noise terms for the first output equation to 0 0 1 0 0 and for the second output equation to 1 0 15 0 0 14 doseTimeNoise A vector of dose time error polynomial coefficients analogous to obsNoise above The default is O for all coefficients This is applied to all dose times regardless of drug so can only be of length 4 e g 800 1 0 1 0 0 15 doseNoise A vector of dose amount error polynomial coefficients analogous to obsNoise above The default is 0 for all coefficients This is applied to all dose amounts regardless of drug so can only be of leng
19. fourth item in the covariate argument list is limits and this functions a bit differently than the limits argument for parameters does 2 above Because the desired limits of a covariate e g weight are generally known unlike for population parameters this argument is specified in the same way as for mean and sd that is a named list with the limits for each covariate If this list item is missing altogether no limits will be applied to simulated covariates If it is supplied then covariates which are not fixed and not included in the list will have the same limits as in the original population If you want to simulate some covariates with limits and some without specify the latter with very large ranges e The fifth and final item in the covariate argument list is fix This is simply a character vector of the names of the covariates that you wish not to simulate i e fix to the values in the template data file Whether you use the means and standard deviations in the population or specify your own the covariance matrix in poppar will be augmented by the covariate covariances The parameter plus covariate means and this augmented covariance matrix will be used for simulations In effect all covariates are moved into the Primary block of the model file to become parameters that are simulated In fact a copy of your model file is made with a c prepended to the model name e g model txt gt c model txt Likewise the data file is
20. npscript bat or itscript bat file in the working directory ITrun and NPrun both return the full path of the output directory to the clipboard By default runs are placed in folders numbered sequentially beginning with 1 Now the output of IT2B or NPAG needs to be loaded into R so the next command does this PMload run number Details of these commands and what is loaded are described in the R documentation PMload and in the following section The run number should be included within the parentheses to be appended to the names of loaded R objects allowing for comparison between runs e g PMload 1 Finally at this point other Pmetrics commands can be added to the script to process the data such as the following Pmetrics User s Guide 18 eC final 1 plot cycle 1 plot op 1 type pop or plot op 1 popl plot op 1 default is to plot posterior predictions for output 1 plot op 1 type pop resid T Of course the full power of R can be used in scripts to analyze data but these simple statements serve as examples If you do not use the PMtree structure we suggest that the R script for a particular project be saved into a folder called Rscript or some other meaningful name in the working directory Folders are not be moved by the batch file Within the script number runs sequentially and use comments liberally to distinguish runs as shown below library Pmetrics R
21. or character vector length 1 e f numeric must correspond to an observation time common to all PMsim objects in simdata rounded to the nearest hour In this case the target statistic will be the ratio of observation at time target type to target This enables testing of a specific timed concentration e g one hour after a dose or C1 which may be called a peak but is not actually the maximum drug concentration Be sure that the time in the simulated data is used e g 122 after a dose given at 120 e time calculate the proportion of the time range specified by start and end time above target The algorithm looks at each pair of concentrations within a simulate profile and if both are below the target the cumulative time is not incremented If both are above the cumulative time is incremented by the time interval between the pair If one is above and the other below the cumulative time is incremented by the fraction above the target as estimated by linear regression between the paired concentrations e auc calculate ratio of area under the curve within the start end time range to target using the trapezoidal approximation in makeAUC to calculate AUC e peak calculate ratio of peak concentration within the start end time range to target e min calculate the ratio of minimum concentration within the start end time range to target 5 success A single value specifying the success statistic e g 0 4 for proportion t
22. row must have an entry 0 observation 1 input e g dose 2 3 are currently unused 4 reset where all compartment values are set to 0 and the time counter is reset to 0 This is useful when an individual has multiple sampling episodes that are widely spaced in time with no new information gathered This is a dose event so dose information needs to be complete This is the elapsed time in decimal hours since the first event It is not currently clock time e g 21 30 although this is planned Every row must have an entry and within a given ID rows must be sorted chronologically earliest to latest Pmetrics User s Guide 8 DUR This is the duration of an infusion in hours If EVID 1 there must be an entry otherwise it is ignored For a bolus i e an oral dose set the value equal to 0 DOSE This is the dose amount If EVID 1 there must be an entry otherwise it is ignored ADDL This specifies the number of additional doses to give at interval II It may be missing for dose events EVID 1 or 4 in which case it is assumed to be 0 It is ignored for observation EVID 0 events Be sure to adjust the time entry for the subsequent row if necessary to account for the extra doses If set to 1 the dose is assumed to be given under steady state conditions ADDL 1 can only be used for the first dose event for a given subject or an EVID 4 event as you cannot suddenly be at steady state in the middle of dosing record unless all
23. simulation are also available for internal model validation the normalized prediction distribution errors NPDE method of Mentr and Escolano 3 and visual predictive checks VPC Both of these can be computed from the same simulation For NPDE the basic idea is that each subject in the population serves as a template for a simulation of 1000 further profiles using the population structural model and parameter values joint probability distribution i e together the population model The simulated profiles are compared to the observed data and the NPDE is generated The command to generate a PMnpde list object is makeNPDE which is documented in R Your current working Pmetrics User s Guide 50 eC directory should be your runs folder as the argument to makeNPDE is the run folder number that you wish to analyze Because of the extensive simulations involved execution of this command can be slow if the population is large the model complex the time horizon long and or the number of observations to be simulated per profile is large There are print and plot methods for NpdeObjects which are contained within PMnpde objects print NpdeObject and plot NpdeObject both of which are also documented within R and much more extensively online at http www biostat fr NPDE index php An example of a NPDE plot is shown in the Plotting section Note that simulation from a population model can be a fickle thing which may lead to er
24. the SIMparse command which is detailed below SIMparse file include exclude combine F silent F Note that SIMparse returns the parsed output of a simulator run as a PMsim object so use the command like this simdata SIMparse sothatsimdata will contain the results The arguments to SIMparse follow 1 file An output file or files of the simulator in the current working directory or the full pathname to the file To load and combine multiple outputs specify files separated by commas or using wild cards For file specification 2 will be matched by just a single numeral or character will be matched by any number of consecutive alphanumeric characters Examples simdata lt SIMparse file simoutl txt simout2 txt simout3 txt simdata SIMparse file simout txt and simdata SIMparse file sim txt All three examples will find the files simout1 txt simout2 txt and simout3 txt in the working directory The second example would also find simout4 txt etc The third example would also find sim 1 txt if that existed 2 include A vector of files to include in the parsing Example simdata lt SIMparse simout txt include c 1 4 If the wildcard match returned four files simouti txt simout2 txt simout3 txt and simout4 txt and you wished to only parse the first and fourth files this would accomplish that filtering 3 exclude A vector of files to exclude in the par
25. values at each cycle normalized to starting values the normalized standard deviation of the population distribution for each parameter at each cycle and the normalized median parameter values at each cycle The default is to omit the first 10 of cycles as a burn in from the plots PMcov makeCov plot PMcov Plots the relationship between any two columns of a PMcov object Pmetrics User s Guide 42 PMmatrix PMreadMatrix plot PMmatrix Plots raw time observation data from a data csv file read by the PMreadMatrix command with a variety of options including joining observations with line segments including doses overlaying plots for all subjects or separating them including individual posterior predictions post objects as described above color coding according to groups and more PMsim SIMparse plot PMsim Plots simulated time concentration profiles overlaid as individual curves or summarized by customizable quantiles e g 5th 25th 50th 75th and 95th percentiles Inclusion of observations in a population can be used to return a visual and numerical predictive check PMnpde makeNPDE plot PMnpde Plots an NPDE qqnorm NPDE histogram NPDE vs time NPDE vs prediction and others to visualize results of simulation based internal model diagnostics accessed with the makeNPDE command More documentation is available at http www biostat fr NPDE index php PMpta makePTA plot PMpta P
26. 0 500 600 700 800 Cycle Cycle Normalized SD Normalized Median 10 1 1 00 0 90 200 300 400 500 600 700 800 Cycle Cycle plot PMcycle object Pmetrics User s Guide 45 Theoretical Quantiles npde 2 Q Q plot versus N 0 1 for npde 3 2 1 0 Sample quantiles npde time Output Frequency 1 2 3 plot PMnpde Time h plot PMsim object Pmetrics User s Guide 46 Sample quantiles npde Predicted out Proportion with T gt MIC of at least 40 0 25 0 5 1 2 4 8 16 32 MIC plot PMpta object Probability of Target Attainment PTA analyses and plots are a powerful application of simulation in Pmetrics Simulated output profiles e g time concentration profiles are automatically compared to a target to generate a pharmacodynamic index PDI This index can be one of several such as the proportion of a dosing interval that the concentration remains above the target the ratio of area under the time concentration curve to target ratio of maximum or minimum concentration to target or ratio of concentration at a specific time to target Here are the details of the makePTA function makePTA simdata simlabels targets target type success outeq 1 free fraction 1 start end 1 simdata A vector of simulator output filenames e g c simout1 txt simout2 txt with wildcard support e g simout or
27. 1 mean id time wt africa ag gender height Ka Ke V Tlagl Pmetrics User s Guide 27 1 60 46 21 160 440395 024616 66 3924 0 554941 2 60 66 30 174 7405 0398 119 476 0 0269964 3 60 46 24 164 899944 0431027 108 649 2 09592 4 60 50 25 165 897547 0564307 119 819 0 688301 5 60 65 22 181 105318 0675052 113 344 0 0186971 6 60 65 23 177 895218 0348829 71 8626 1 99784 7 60 51 27 161 215198 0832836 35 2243 1 79653 8 60 51 22 163 895481 0348882 71 847 99849 9 60 55 23 174 789913 0439419 101 783 0 879688 10 60 52 32 163 655786 0615878 61 6927 0 801376 11 60 56 34 165 583223 068323 73 1082 1 33855 12 60 47 54 160 470306 0306883 91 8595 1 02535 13 60 60 24 180 215198 0832837 35 2243 1 79559 14 60 59 26 174 579989 0439032 117 837 0 336257 15 60 43 19 150 795628 034209 72 2038 1 05869 16 60 64 25 173 752955 0352986 89 6704 0 687976 17 60 54 23 170 891255 0734126 63 3196 1 17994 18 60 44 20 164 894613 023192 75 9273 1 76035 19 60 50 36 168 662597 0621169 30 9852 1 92418 20 60 59 31 170 215198 0832837 35 2243 1 79584 PMfinal summary x lower 0 025 upper 0 975 summary PMfinal For NPAG runs this function will generate a data frame with weighted medians as central tendencies of the population points with an upper lower default 9596 confidence interval 9596 CI around the median and the median absolute weighted deviation MAWD from the median as a measure of the variance with its 9596
28. 2B run number and output equation this function will iterate through the data csv file using each subject as a template to simulate nsim new individuals from the population prior It is HIGHLY recommended to use the default value of 1000 for nsim for the most valid calculation of NPDE More than this could take a long time to execute The mean population values will be used for each parameter and the covariance matrix Errors may arise if extreme or negative concentrations are simulated from excessively large covariance matrices Because considerable time may be necessary to make the NpdeObject it will be added as an NPDE item to the NPAGout Rdata or IT2Bout Rdata objects so that it will be loaded the next time PMload is run Additionally the combined simulations for all the subjects in the dataset will be saved as a sim item in the NPAGout Rdata or IT2B Rdata objects Summarizing Pmetrics Objects There are summary commands available for several Pmetrics objects as detailed below All objects can be summarized by the R command summary x where x is the object you wish to summarize PMcov summary x icen median summary PMcov Summarize a PMcov object by creating a data frame with each subject s covariate values and Bayesian posterior parameter values summarized according to icen Default is median covariate values and Bayesian posterior parameter values but could be mean Example data PMex1 summary cov
29. ANT The order capitalization and names of the header and the first 12 columns are fixed All entries must be numeric with the exception of ID and for non required placeholder entries POPDATA DEC 11 ID EVID TIME DUR DOSE ADDL II INPUT OUT OUTEO CO C1 C2 C3 GOV GH 1 0 0 400 1 i GH 0 0 5 0 42 1 0 01 0 1 0 0 GH 0 1 0 46 1 0 01 0 1 0 0 GH 0 2 i 247 1 0 01 0 1 0 0 GH 4 0 0 150 1 i GH 1 35 05 150 1 0 01 0 1 0 0 GH 0 9 12 055 1 0 010 1 0 0 GH 0 24 j 052 1 0 010 1 0 0 1423 1 0 1 400 1 12 1 1423 1 01 0 100 2 i i 1423 0 1 99 1 0 01 0 1 0 0 1423 0 2 0 38 1 0 01 0 1 0 0 1423 0 2 16 2 0 05 0 2 0 11 0 002 POPDATA DEC 11 This is the fixed header for the file and must be in the first line It identifies the version It is not the date of your data file ZID EVID TIME This field must be preceded by the symbol to confirm that this is the header row It can be numeric or character and identifies each individual All rows must contain an ID and all records from one individual must be contiguous Any subsequent row that begins with will be ignored which is helpful if you want to exclude data from the analysis but preserve the integrity of the original dataset or to add comment lines IDs should be 11 characters or less but may be any alphanumeric combination There can be at most 800 subjects per run This is the event ID field It can be 0 1 or 4 Every
30. NM2PM in R Pmetrics User s Guide 9 Model Files Model files for Pmetrics are ultimately Fortran text files with a header version of TSMULT As of Pmetrics version 0 30 we have adopted a very simple user format that Pmetrics will use to generate the Fortran code automatically for you Version 0 4 additionally eliminates the previously separate instruction file A model library is available on our website at http www lapk org pmetrics php Naming your model files The default model file name is model txt but you can call them whatever you wish However please keep the number of characters in the model file name s 8 When you use a model file in NPrun ITrun ERRrun or SIMrun Pmetrics will make a Fortran model file of the same name temporarily renaming your file At the end of the run your original model file will be in the inputs subfolder of the run folder and the generated Fortran model file will be called model for and moved to the etc subfolder of the run folder If your model is called mymodel txt then the Fortran file will be mymodel for You can still use appropriate Fortran model files directly but we suggest you keep the for extension for all Fortran files to avoid confusion with the new format If you use a for file as your model you will have to specify its name explicitly in the NPrun ITrun ERRrun or SIMrun command since the default model name again is model txt
31. Pmetrics User Manual July 2015 An R package for parametric and non parametric modeling and simulation of pharmacokinetic and pharmacodynamic systems Se USC University of LAPKB WY Southern California Laboratory of Applied Pharmacokinetics amp Bioinformatics Table of Contents PATE OCU CTO WN e 3 iersbi dudii2adi cNee C X 3 BIECIEITUT PIE EE 3 System Requirements and Installation ee eee eee eee eene eene eee eene eene nn nennen nennen nnne e sensn na sonne n ens ss sensere 3 What This Manual S D MM 5 cinia n ircmM E 5 Pmetrics Components teo cee nee e oet eter ope ee Ie pne appe eU oo ves ees eap a Use eve ee eb covectesvecccbwestoautedccdGadevcodvencuavoncctcawes 5 te TES orrinrauiaidezen i tee 6 G neral WorkflOW E 7 Pmetrics Inp t dl eL 8 Data csu ad nce 8 Manipulation of CSV Tiles ici e 9 DDPOIIBJIIf 10 How to use R and PENS cca is
32. al from 72 to 96 hours with concentration target pta l1 lt makePTA simdata simlist targets c 0 25 0 5 1 2 4 8 16 target type time success 0 6 start 72 end 96 Example 2 parse the results into a list simlist SIMparse simout txt Pmetrics User s Guide 48 eC make the PTA with sampled targets using a file with the distribution of vancomycin MICs for staphylococcus aureus and define success as an AUC target 2 400 from 72 to 96 hours pta 1 lt makePTA simdata simlist targets makePTAtarget staph csv target type auc success 400 start 72 end 96 Example 3 parse the results into a list simlist SIMparse simout txt make the PTA with a target concentration of 4 mg L 12 hours after e g time 84 a steady state dose given every 12 hours pta 1 lt makePTA simdata simlist targets 4 target type 84 success 1 The output of makePTA isalistof class PMpta which has 2 objects 1 results A data frame with the following columns e simnumis the number of the simulation e idis the simulated profile number within each simulation e target is the specified target e pdiis the target pharmacodynamic index e g time gt target auc target etc 2 outcome A data frame summarizing the results with the following columns e simnum and target are as for results If targets was specified via makePTAtarget to be a sampled distribution then the target c
33. ard deviation error of the observation based on the assay error polynomial final class PMfinal list popPoints NPAG only Data frame of the final cycle joint population density of grid points with column names equal to the name of each random parameter plus prob for the associated probability of that point popMean The final cycle mean for each random parameter distribution popSD The final cycle standard deviation for each random parameter distribution popCV The final cycle coefficient of variation for each random parameter distribution popVar The final cycle variance for each random parameter distribution popCov The final cycle covariance matrix for each random parameter distribution popCor The final cycle correlation matrix for each random parameter distribution popMedian The final cycle median for each random parameter distribution gridpts NPAG only The initial number of support points ab Matrix of boundaries for random parameter values For NPAG this is specified by the user prior to the run for IT2B it is calculated as a user specified multiple of the SD for the parameter value distribution ul ilu i Pmetrics User s Guide 20 postPoints NPAG only Data frame of the Bayesian posterior parameter points for each of the first 100 subjects with the following columns id subject ID point point number for that subject parameters parameters in the model prob probability of each point in th
34. ated as fixed parameters with user specified values It should be a list of variable names one name to a line Variable names should be 11 characters or fewer Some variable names are reserved for use by Pmetrics and cannot be used as primary variable names The number of primary variables must be between 2 and 32 with at most 30 random or 20 fixed On each row following the variable name include the range for the parameter that defines the search space These ranges behave slightly differently for NPAG IT2B and the simulator Pmetrics User s Guide 10 eC e For all engines the format of the limits is min max A single value will fix that parameter to the specified value e For NPAG the limits are absolute i e the algorithm will not search outside this range e For IT2B the range defines the Bayesian prior distribution of the parameter values for cycle 1 For each parameter the mean of the Bayesian prior distribution is taken as the middle of the range and the standard deviation is xsig range see IT2B runs Adding an exclamation point to a line will prevent that parameter from being assigned negative values NPAG and the simulator will ignore the pluses as the ranges are absolute for these engines e The simulator will ignore the ranges with the default value of NULL for the limits argument If the simulator limits argument is set to NA which will mean that these ranges will be used as the limits to truncate the simulation
35. ax Maximum predicted concentration after the first dose tmax Time to cmax cl Clearance calculated as dose aucinf vdss Volume of distribution at steady state calculated as cl mrt thalf Half life of elimination calculated as In 2 k dose First dose amount for each subject makeErrorPoly This function plots first second and third order makeErrorPoly polynomial functions fitted to pairs of observations and associated standard deviations for a given output assay In this way the standard deviation associated with any observation may be calculated and used to appropriately weight that observation in the model building process Observations are weighted by the reciprocal of the variance or squared standard deviation Output of the function is a plot of the measured observations and fitted polynomial curves and a list with the first second and third order coefficients Pmetrics User s Guide 25 makePTA This function performs a Probability of Target makePTA Attainment analysis for a set of simulated doses and time concentration profiles Targets e g Minimum Inhibitory Concentrations the type of target attainment i e time above target Cmax target AUC target Cmin target or Cx target where x is any time point and the success threshold e g time gt 0 7 or Cmax target gt 10 can all be specified Output is a list class PMpta with two objects e Results A data frame with the following columns simnum is the
36. b and download our products including Pmetrics Pmetrics User s Guide 3 eC Pmetrics is distributed as a package source file archive tgz for Mac zip for Windows tar gz for Linux Do not open the archive To install Pmetrics from the R console use the command install packages file choose and navigate when prompted to the folder in which you placed the Pmetrics package archive zip or tgz file Pmetrics will need the following R packages for some functions chron Defaults and RZ2HTML However you do not have to install these if you do not already have them in your R library They should automatically be downloaded and installed when you use PMbuild or the first time you use a Pmetrics function that requires them but if something goes awry such as no internet connection or busy server you can do this manually Fortran In order to run Pmetrics a Fortran compiler is required After you have installed Pmetrics the first command to run is PMbuild which will verify an installed fortran compiler on your system or link you to the OS specific page of our website with explicit instructions and a link to download and install gfortran on your system Details of this procedure follow but are not relevant if you already have a compiler installed For Mac users the correct version of gfortran will be downloaded for your system Mavericks 64 bit Mountain Lion 64 bit Lion 64 bit Snow Leopard 64 or 32 bit You will also be provided
37. ched for viewing of the HTML report page It will also generate a tex file suitable for processing by a IATEX engine to generate a pdf report See the Pmetrics Outputs section Within Pmetrics there are also functions to manipulate data csv files and process and plot extracted data Manipulate data csv files PMreadMatrix PMcheck PMwriteMatrix PMmatrixRelTime PMwrk2csv NM2PM Process data makeAUC makeCov makeCycle makeFinal makeOP makeNCA makePTA makeErrorPoly Plot data plot PMcov plot PMcycle plot PMfinal plot PMmatrix plot PMop plot PMsim plot PMdiag plot PMpta Model selection and diagnostics PMcompare plot PMop with residual option makeNPDE PMstep Pmetrics function defaults PMwriteDefaults Again all functions have extensive help files and examples which can be examined in R by using the help command or command syntax Customizing Pmetrics Options You can change global options in Pmetrics setPMoptions sep dec Currently you can change two options sep will allow Pmetrics to read data files whose field separators are semicolons and decimal separators are commas e g setPMoptions sep dec These options will persist from session to session until changed getPMoptions will return the current options Pmetrics User s Guide 6 eC General Workflow The general Pmetrics workflow for IT2B and NPAG is shown in the following diagram You supply these files Pmetrics does the re
38. del comparison To compare models with PMcompare simply enter two or more PMetrics data objects e g PMcompare NPdata 1 NPdata 2 NPdata 3 These should be of the NPAG or IT2B class made either by using PMload or NPparse ITparse Although itis possible to compare models of mixed classes the validity of this is dubious The return object will be a data frame with summaries of each model and key metrics such as log likelihood final cycle Akaike Information and Bayesian Information Criteria bias and imprecision of predictions relative to observations from the population prior distribution and individual posterior distributions and a p value for model The p values are a comparison of the joint distribution of population parameter values for all models using the first model in the list as the reference Only parameters with common names to all models are included Data objects supplied to PMcompare should not be zero cycle runs which will generate an error message By specifying the option plot T observed vs predicted plots for all the models will be generated The comparison is based on the nearest neighbors approach in the MTSKNN package The option to generate residual plots of prediction errors described next can be specified with the additional switch resid T which is ignored if plot F Model bias and imprecision In both PMcompare and in plots of PMop objects with reg T the default Pmetrics supplies two statistics bias
39. e posterior for each patient cycle class PMcycle list Vector Vector of names of the random parameters names of the random Vector of names of the random parameters Matrix of cycle number and 2 Log likelihood at each cycle gamlam A matrix of cycle number and gamma or lambda at each cycle see item 16 under NPAG Runs below for a discussion of gamma and lambda A matrix of cycle number and the mean of each random parameter at each cycle normalized to initial mean A matrix of cycle number and the standard deviation of each random parameter at each cycle normalized to initial standard deviation median A matrix of cycle number and the median of each random parameter at each cycle normalized to initial standard deviation aic A matrix of cycle number and Akaike Information Criterion at each cycle bic A matrix of cycle number and Bayesian Schwartz Information Criterion at each cycle cov class PMcov Subject identification data Pf time OS Time for each covariate Time foreach covariate entry covariates Covariate values for each subject at each time extracted from the raw data file Pmetrics User s Guide 21 parameters Mean median or mode of Bayesian posterior distribution for each random parameter in the model Mode summaries are available for NPAG output only and the default is median Values are recycled for each row within a given subject with the number of rows driven
40. e values in the template data file used for simulations 11 seed The seed for the random number generator For nsub gt 1 should be a vector of length equal to nsub Shorter vectors will be recycled as necessary Default is 17 Examples SIMrun seed c 2 3 This will run the simulator for the two subjects in the data file with 2 as the seed for the first subject and 3 for the second subject The value of the seed is that it will generate the same random parameter sets each time the seed is specified This allows random but reproducible results for comparing simulations where other factors have changed such as the dosing regimen used in the template data file Setting the seed to fewer than the number of subjects e g 1 seed for 3 subjects will result in each subject having the same set of simulated parameter values because Pmetrics recycles the seed as needed This may or may not be desirable 12 ode Ordinary Differential Equation solver log tolerance or stiffness Default is 4 ie 0 0001 Higher values will result in faster simulations but simulated concentrations may not be as accurate This is the same parameter as for NPAG and IT2B runs 13 obsNoise This is the noise added to simulated observations If present will override any other values in the data file or model file Should be a vector of length 4 times the number of output equations e g c 0 1 0 1 0 0 for one output and c 0 1 0 1 0 0 0 01 0 2 0 001 0
41. ean squared error RMSE 0 8916 Percent root mean squared error RMSE 12 3141 Mean weighed squared prediction error 0 992 Bias adjusted mean squared prediction error 0 7655 Bias adjusted mean weighted squared prediction error imprecision 0 9813 This function also returns a list with three items The first item of the list is a data frame with the minimum first quartile median third quartile maximum mean and standard deviation for times observations and predictions in x The second is a data frame of one row whose columns contain the mean prediction error the mean weighted prediction error bias the mean squared prediction error root mean sqaured error RMSE percent root mean squared error squared prediction error the bias adjusted mean squared prediction error and the bias adjusted mean weighted squared prediction error imprecision The third is itself a list of 6 items related to the weighted prediction bias which are the mean the standard error of the mean the 9596 confidence interval the t statistic the degrees of freedom and the p value compared to a bias of 0 In the example above the p value is 0 9718 PMpta summary x ci 0 95 summary PMpta This function returns a list with two named objects pta probability of target attainment and pdi pharmacodynamic index pta A data frame with the following columns simnum target prop success pdi mean and pdi sd simnum is t
42. ed by the command PMpatch if available Note that patches must be reinstalled with this command every time you launch Pmetrics until the bug is corrected in the next version As of version 1 0 0 Pmetrics has graphical user interface GUI capability for many functions Using PMcode function will launch the GUI in your default browser While you are interacting with the GUI R is listening and no other activity is possible The GUI is designed to generate Pmetrics R code in response to your input in a friendly intuitive environment That code can be copied and pasted into your Pmetrics R script You can also see live plot previews with the GUI All this is made possible with the shiny package for R Currently the following GUIs are available PMcode NPrun PMcode ITrun PMcode plot More are coming Pmetrics Components There are three main software programs that Pmetrics controls IT2B is the ITerative 2 stage Bayesian parametric population PK modeling program It is generally used to estimate parameter ranges to pass to NPAG It will estimate values for population model parameters under the assumption that the underlying distributions of those values are normal or transformed to normal e NPAG is the Non parametric Adaptive Grid software It will create a non parametric population model consisting of discrete support points each with a set of estimates for all parameters in the model plus an associated probabili
43. er s Guide 14 b npl numeqt ndrug nadd rateiv cv n nd ni nup nuic np nbcomp psym fa tlag tin tout array of primary parameters input rates input boluses internal output equation number input number covariate number intravenous input for inputs when DUR gt 0 in data files covariate values array number of compartments internal internal internal internal number of primary parameters bolus compartment array names of primary parameters biovailability lag time internal internal ul ilu i Pmetrics User s Guide 15 Complete Example Here is a complete example of a model file as of Pmetrics version 0 40 and higher Notes By omitting a Diffeq block with ODEs Pmetrics understands that you are specifying the model to be solved algebraically In this case at least KE and V must be in the Primary or Secondary variables KA KCP and KPC are optional and specify absorption and transfer to and from the central to a peripheral compartment respectively The comment line C this weight is in kg will be ignored Pmetrics User s Guide 16 Brief Fortran Tutorial Much more detailed help is available from http www cs mtu edu shene COURSES cs201 NOTES fortran html E emm m me LLL ENNIUS IF logical expression one statement IF T gt 100 CL 10 IF logical expression THEN IF T gt 100 THEN statements CL 10 END IF V 10 END IF IF logical express
44. etrics a Nonparametric and Parametric Pharmacometric Modeling and Simulation Package for R Ther Drug Monit 2012 34 467 476 Disclaimer You the user assume all responsibility for acting on the results obtained from Pmetrics The USC Laboratory of Applied Pharmacokinetics members and consultants to the Laboratory of Applied Pharmacokinetics and the University of Southern California and its employees assume no liability whatsoever Your use of the package constitutes your agreement to this provision System Requirements and Installation Pmetrics and all required components will run under Mac Unix Windows and Linux There are three required software components which must be installed on your system in this order 1 The statistical programming language and environment R 2 The Pmetrics package for R 3 gfortran or some other Fortran compiler A fourth highly recommended but optional component is Rstudio a user friendly wrapper for R It can be installed any time after installing R i e step 1 above All components have versions for Mac Windows and Linux environments and 32 and 64 bit processors All are free of charge R R is a free software environment for statistical computing and graphics which can be obtained from http www R project org Pmetrics is a library for R Pmetrics If you are reading this manual then you have likely visited our website at http www lapk org where you can select the software ta
45. ets may and likely will be different than those specified by poppar Examples SIMrun limits NULL SIMrun limits NA SIMrun limits 3 SIMrun limits c 0 5 SIMrun limits matrix c 0 1 0 5 1000 nrow 2 byrow T The first example has no limits The second example uses the limits on the primary parameters in the model file The third example uses the lower limits on the primary parameters in the model file and multiplies the upper limits of each primary parameter in the model file by 3 The fourth example multiplies the lower limits of the primary parameter values in the model file by 0 and the upper limits are multiplied by 5 The fifth example sets a custom matrix of limits with ranges of 0 to 1 for the first parameter and 0 5 to 1000 for the second 3 model Name of a suitable model file template in the working directory The default is model txt This file will be converted to a fortran model file If it is detected to already be a fortran file then the simulation will proceed without any further file conversion This is the same model file used for NPAG and IT2B runs 4 data Either a PMmatrix object previously loaded with PMreadMatrix or character vector with the filename of a Pmetrics data file that contains template regimens and observation times This file is the same format as for NPAG and IT2B runs and the value for the OUT column EVID 0 rows in this file can be coded as any number s other tha
46. filename and creates a PMmatrix object in memory which can be plotted see plot PMmatrix or otherwise analyzed PMcheck PMmatrix filename model This function will check a csv file named filename or a PMmatrix data frame containing a previously loaded csv file the output of PMreadMatrix for errors which would cause the analysis to fail If a model file is provided and the data file has no errors it will also check the model file for errors See PMcheck for details in R PMwriteMatrix data frame filename This function writes an appropriate data frame as a new csv file It will first check the data frame for errors via the PMcheck function above and writing will fail if errors are detected This can be overridden with override T PMmatrixRelTime This function converts dates and clock times of specified formats into relative times for use in the NPAG IT2B and Simulator engines See PMmatrixRelTime for details PMwrk2csv This function will convert old style single drug USC PACK wrk formatted files into Pmetrics data csv files Details are available with PMwrk2csv in R NM2PM Although the structure of Pmetrics data files are similar to NONMEM there are some differences This function attempts to automatically convert to Pmetrics format It has been tested on several examples but there are probably NONMEM files which will cause it to crash Running PMcheck afterwards is a good idea Details can be found with
47. found in the data file If none are present or if the model declaration line contains an exclamation point the values here will be used Example 1 estimated lambda starting at 0 4 one output use data file coefficients but if missing use 0 1 0 1 0 0 Err L 0 4 0 1 0 1 0 0 Example 2 fixed gamma of 2 two outputs use data file coefficients but if missing use 0 1 0 1 0 0 for the first output but use 0 3 0 1 0 0 for output 2 regardless of what is in the data file Err G 2 0 1 0 1 0 0 0 3 0 1 0 0 Extra This block is for advanced Fortran programmers only Occasionally for very complex models additional Fortran subroutines are required They can be placed here The code must specify complete Fortran subroutines which can be called from other blocks with appropriate call functions As stated earlier sometimes it is important to preserve spacing and formatting in Fortran code that you might insert into blocks particularly the EXTRA block If you wish to do this insert format and format before and after any code that you wish to reproduce verbatim with spacing in the fortran model file Reserved Names The following cannot be used as primary covariate or secondary variable names They can be used in equations however Reserved Variable Function in Pmetrics ndim internal t time X array of compartment amounts xp array of first derivative of compartment amounts rpar internal ipar internal Pmetrics Us
48. he number of the simulation target is the specified target success has the proportion with a ratio prop success pdi mean and pdi sd are the mean and standard deviation of the pharmacodynamic index e g AUC MIC for each simulation and target pdi A data frame with the following columns target simnum lowerCI median upperCI target and simnum are as above lowerCI median and upperCl are the lower limit median and upper limit of the confidence interval for the pdi whose width is specified by ci Pmetrics User s Guide 32 eC NPAG Runs Here we provide details of the arguments available to the NPrun command You must have a data and model file in your working directory In the R syntax below any argument with a value has a default equal to that value and if you wish to use that default you do not have to specify the argument in your function call to NPrun NPrun model model txt data data csv run include exclude ode 4 tol 0 01 salt cycles 100 indpts icen median aucint idelta 12 prior overwrite F nocheck F parallel NA 1 model Name of a suitable model file template in the working directory or an existing previous run number corresponding to a folder in the current working directory that used the same model file as will be used in the current run If this is supplied then the model file will be copied into the current working directory for convenience If not supplied the defa
49. ime start to end above target or 100 for peak target 6 outeq An integer specifying the number of the simulated output equation to use Default is 1 free fraction Proportion of free active drug from 0 to 1 Default is 1 i e 100 free drug or 0 protein binding 8 start Specify the time to begin PTA calculations Default is a vector with the first observation time for subjects in each element of simdata e g dose regimen If specified as a vector values will be recycled as necessary 9 end Specify the time to end PTA calculations so that PTA is calculated from start to end Default for end is the maximum observation time for subjects in each element of simdata e g dose regimen If specified as a vector values will be recycled as necessary Subjects with insufficient data fewer than 5 simulated observations for a specified interval will trigger a warning Ideally then the simulated dataset should contain sufficient observations within the interval specified by start and end If a simulation exists such that simout1 txt simout4 txt exist in the current working directory and each contain a different simulated dosing regimen of 100 mg daily 50 mg bid twice daily 200 mg daily or 100 mg bid all given for 5 days then here are examples of different PTAs Example 1 parse the results into a list simlist SIMparse simout txt make the PTA with discrete targets and define success as proportion 2 0 6 of interv
50. ion THEN IF T gt 100 THEN statements 1 CL 10 ELSE ELSE statements 2 CL CL END IF END IF Pmetrics User s Guide 17 How to use R and Pmetrics Setting up a Pmetrics project When beginning a new modeling project it is convenient to use the command PMtree project name This command will set up a new directory in the current working directory named whatever you have included as the project name For example a directory called DrugX will be created by PMtree DrugX Beneath this directory several subdirectories will be also created Rscript Runs Sim and src The Rscript subdirectory will contain a skeleton R script to begin Pmetrics runs in the new project The Runs subdirectory should contain all files required for a run described next and it will also contain the resulting numerically ordered run directories created after each Pmetrics NPAG or IT2B run The Sim subdirectory can contain any files related to simulations and the src subdirectory should contain original and manipulated source data files Of course you are free to edit this directory tree structure as you please or make your own entirely Getting the required files to run Pmetrics When you wish to execute a Pmetrics run you must ensure that appropriate Pmetrics model txt and data csv files are in the working directory i e the Runs subdirectory of the project directory R can be used to help prepare the data csv file by importing and man
51. ipulating spreadsheets e g read csv The Pmetrics function PMcheck can be used to check a csv file or an R dataframe that is to be saved as a Pmetrics data csv file for errors It can also check a model file for errors in the context of a datafile e g covariates that do not match PMcheck fix T attempts to automatically rid data files of errors The function PMwriteMatrix can be used to write the R data object in the correct format for use by IT2B NPAG or the Simulator You can also download sample data and scripts from the Pmetrics downloads section of our website Edit prior versions of model files to make new model files Using scripts to control Pmetrics As you will see in the skeleton R script made by PMtree and placed in the Rscript subdirectory if this is a first time run the R commands to run IT2B or NPAG are as follows Recall that the character is a comment character library Pmetrics Run 1 add your run description here setwd working directory NPrun for NPAG or ITrun for IT2B The first line will load the Pmetrics library of functions The second line sets the working directory to the specified path The third line generates the batch file to run NPAG or IT2B and saves it to the working directory NOTE On Mac systems the batch file will be automatically launched in a Terminal window On Windows systems the batch file must be launched manually by double clicking the
52. is 1000 predInt The interval in fractional hours for simulated predicted outputs at times other than those specified in the template data The default is 0 which means there will be simulated outputs only at times specified in the data file Values of predInt 0 result in simulated outputs at the specified value of predInt e g every 15 minutes for predInt 0 25 from time 0 up to the maximal time in the template file per subject if nsub gt 1 You may also specify predInt as a vector of 3 values e g c 1 4 1 similar to the R command seq where the first value is the start time the second is the stop time and the third is the step value Outputs for times specified in the template file will also be simulated To simulate outputs only at the output times in the template data i e EVID 0 events use predInt 0 which is the default Note that the maximum number of predictions total is 594 so the interval must be sufficiently large to accommodate this for a given number of output equations and total time to simulate over If predInt is set so that this cap is exceeded predictions will be truncated Examples SIMrun SIMrun predInt 1 SIMrun predInt c 1 12 0 5 The first example is the same as predInt 0 the default which will simulate observations only at the times specified in the template date file i e the times of all EVID 0 events The second example simulates observations hourly starting at 1 hour plus an
53. is se casa an esc rea COE UL FERE REUR scapes Min REESE GUDDE NUS dpa dE eid Dos Duc OE ebere 18 Piietrics Data Ob BCIS iion reir neges EX ERE ELK Dra ER wk I s UeFE C a prU XR RNMP E om T 19 Making New Pmetrics Objects oe onera bobina COE Lpbi ao AP GEI aX ETa VPE TERRIER DARE DL OFF NENNT EE KK Fr DoNRN Loren UE 24 Summarizing Pmetrics Ob jGCts rose noid irre bv vw csv e Dvkhik er rbd dedica kh ded bell x i ad viru xw V Pa kiss KE EVK I DO OO 27 PIVICOY 27 Leu TT 28 11 PHA 31 PIV PUG ssavccesicassszascensassscsesccodsdsessssessceasssssussdensssasnacsaniscdotssdnesdanseseoss daesso0sssebesssaesseassdsesscucssseasssesessossscsasasssentees 32 dcin 33 HT 2B RUNS E T 35 DNO 35 Simulator RUNS t 36 biDrinp p eT Tc NEVO VK KN 42 I3 5 hd Club cR 44 Probability of Target Attala HE oos sies ria eux Dex ops Do te ERE REVx S bREERI RC FVEPEG RIS GEM KEE E PRKKEEEPP eu RE REV OPE AME PER VER EEEK RNE KE 47 Model DiagnoSLits uui ap puerta FARER Y edv E IAP E Erb VE REREPEUVR EE auena
54. kj ze E H HE FOLLOWING ARE MEAN SD MIN TO MAX INPUTS Number of doses per subject input 1 6 000 0 000 6 000 to 6 000 Dose per subject input 1 585 000 45 189 450 000 to 600 000 OUTPUTS Number of obs per subject outeq 1 6 950 0 224 6 000 to 7 000 Observation per subject outeq 1 7 241 3 799 1 860 to 20 150 COVARIATES wt 541 538 17 173 43 000 to 66 500 africa 1 000 0 000 1 000 to 1 000 ages 27 035 7 717 19 000 te 54 000 gender 0 749 0 434 0 000 to 1 000 height 167 792 7 562 150 000 to 161 000 Note See help summary PMmatrix for accessing specific items by name An object of class summary PMmatrix is also returned This object is a list with the following items nsub The number of subjects ndrug The number of drug inputs numeqt The number of output equations nobsXouteq The number of observations by output equation missObsXouteq The number of missing observations by output equation ncov The number of covariates covnames The covariate names ndoseXid The number of doses per input per subject nobsXid The number of observations per output equation per subject doseXid The doses per input per subject obsXid The observations per output per subject formula The results of including a formula To include a formula use standard R notation for the formula and specify the aggregating function with FUN Example To calculate average do
55. lot like the one below Although the template subjects all received only one dose of drug in this interval they were not sampled at the same times so when the simulations were combined it leads Output Time h to the jagged quantiles This can be improved i e smooth the quantile curve by setting the binSize argument when plotting which controls the width of binning interval for simulated concentrations in time units e g hours For example a binSize of 0 5 will pull all simulated concentrations 0 5 hours into the same time bin The default is 0 i e no binning The return object when obs is included is a list with the following items 1 npc A dataframe with three columns quantile prop less pval quantile are those specified by theprob argument to the plot call prop less are the proportion of simulated observations at all times less than the quantile pval is the P value of the difference in the prop less and quantile by the beta binomial test 2 simsum A dataframe with the quantile concentration at each simulated time with lower and upper confidence intervals 3 obs A dataframe similar to an PMop object made by makeOP with the addition of the quantile for each observation Pmetrics User s Guide 51 eC So to get the NPC simply type from the example above vpcSnpc Visual predictive checks can also be made from a PMnpde object made with the makeNPDE function with the following syntax plot npde
56. lots superimposed curves corresponding to each dose with target e g MIC on the x axis and proportion of the simulated time concentration profiles for the dose with a target statistic e g time gt MIC above a user defined success threshold Pmetrics User s Guide 43 Examples of Pmetrics plots Time h Predicted 6 R squared 0 949 Inter 0 2 95 Cl 0 261 to 0 662 Slope 0 966 95 CI 0 885 to 1 05 Bias 0 0315 Imprecision 0 346 plot PMop object D Agostino P 0 267 2 2 Shapiro Wilk P 0 008 Kolmogorov Smimoff HP 0 E o 4 e o o 3 EE 8 t So g 8 i 3 H M e 3 3 82 3 P i i o4 t LR I 1 i iTi o E a 5 i 8 EI b 9 o o T LJ b Medh 0 27 P 0 169 SD 2 34 o 10 5 E E 2 e 90 ns M0 ws s o 5 Predicted Tine Weighted residual error plot PMop object resid T 0 05 0 06 0 07 0 08 0 09 0 10 0 05 0 06 0 07 0 08 009 0 10 0 05 0 06 0 07 008 0 09 0 10 e s 8 S E S 5 S 8 E 8 S 2 Tlag plot PMfinal object Pmetrics User s Guide 44 2 x Log likelihood AIC BIC B E S Bc F AIC uw e N e o 8 8 N oor 200 300 400 500 600 700 800 200 300 400 500 600 700 800 Cycle Cycle Gamma Lambda Normalized Mean e 5 d 8 N B S Gel 8 E 200 300 400 500 600 700 800 200 300 40
57. mulator is a Fortran executable compiled and run in an OS shell It is documented with an example within R You can access this by usingthe help SIMrun or SIMrun commands from R In order to complete a simulator run you must include a data csv file and a model file in the working directory The structure of these files is identical to those used by NPAG and IT2B The data csv contains the template dosing and observation history as well as any covariates Observation values the OUT column for EVID 0 events can be any number they will be replaced with the simulated values However do not use 99 as this will simulate a missing value which might be useful if you are testing the effects of such values A good choice for the OUT value in the simulator template csv file is 1 You can have any number of subject records within a data csv file each with its own covariates if applicable Each subject will cause the simulator to run one time generating as many simulated profiles as you specify from each template subject This is controlled from the SIMrun command with the include and nsimarguments The first specifies which subjects in the data csv file will serve as templates for simulation The second specifies how many profiles are to be generated from each included subject Pmetrics User s Guide 36 eC Simulation from a non parametric prior distribution from NPAG can be done in one of two ways The first is simply to take the mean
58. n 99 We typically use 1 The number s will be replaced in the simulator output with the simulated values 5 split Boolean operator controlling whether to split an NPAG PMfinal object into one distribution per support point with means equal to the vector of parameter values for that point and covariance equal to the population covariance divided by the number of support points See the figures below to better understand what happens when split is true or false The red points are the non parametric population distribution while the gold wireframe overlays are smoothed probability functions from which samples are drawn in simulations split F unimodal multivariate sampling split T multimodal multivariate sampling Pmetrics User s Guide 38 10 include A vector of subject IDs in the data file to iterate through with each subject serving as the source of an independent simulation If missing all subjects in the datafile will be used This is analogous to the same argumentin NPrun and ITrun exclude A vector of subject IDs to exclude in the simulation e g c 4 6 14 16 20 If a makecsv filename is supplied ID numbers will be of the form nsub nsim e g 1 001 through 1 1 for the first subject 2 001 through 2 1 for the second subject etc if 1000 simulations are made from each subject This is analogous to the same argumentin NPrun and ITrun nsim The number of simulated profiles to create per subject Default
59. n a folder labeled 2 If you haven t already deleted the old run folder you can set overwrite 15 below to True 4 include Vector of subject id values in the data file to include in the analysis The default missing is all Examples NPrun include c 1 3 5 Include only subjects with ID numbers of 1 2 3 and 5 5 exclude Vector of subject id values in the data file to exclude in the analysis The default missing is none Be careful if you include and exclude in the same run as you may have conflicting statements Examples NPrun exclude c 1 3 5 Exclude subjects with ID numbers of 1 2 3 and 5 6 ode Ordinary Differential Equation solver logio tolerance or stiffness Default is 4 i e 0 0001 Higher values will result in faster runs but parameter estimates may not be as accurate It is ignored if the model does not use differential equations Examples NPrun ode 3 Do the run with a stiffness of 0 001 which will be faster than the default but perhaps less accurate 7 tol Tolerance for convergence of NPAG Smaller numbers make it harder to converge Default value is 0 01 Examples NPrun tol 0 001 Do the run with a tolerance of 0 001 which will be slower than the default but perhaps more accurate 8 salt Vector of salt fractions for each ndrug default is 1 for each drug This is not the same as bioavailability Examples NPrun salt c 1 0 9 This sets the salt fraction for the first drug to 1 and for the second drug
60. nd You must have a data and model file in your working directory In the R syntax below any argument with a value has a default equal to that value and if you wish to use that default you do not have to specify the argument in your function call to ITrun ITrun model model txt data data csv run include exclude ode 4 salt cycles 100 tol 0 001 xdev 5 icen median overwrite F nocheck F model Same as for NPAG data Same as for NPAG run Same as for NPAG include Same as for NPAG exclude Same as for NPAG ode Same as for NPAG salt Same as for NPAG O NO DW PWN P cycles Maximum number of IT2B cycles to run Default is 100 but the value can be 1 or greater Note that unlike NPAG you cannot enter a value of 0 If you enter an integer greater than 0 the engine will terminate at convergence or the number of cycles you specify whichever comes first Examples ITrun cycles 1000 The first example will allow IT2B to run 1000 cycles before terminating 9 tol Same as for NPAG except the default value is 0 001 10 xdev Multiple of standard deviations for parameters to be used in NPAG as a range Default is 5 The ranges can be found in the FROMO0001 file in the outputs folder of the run folder IT2B is primarily used to estimate ranges for parameter values to be passed to NPAG Because it is parametric it is not constrained to the initial parameter ranges in the model file unlike NPAG Exam
61. number of outputs and times of observations also may differ although combining these may lead to strange plots since not all profiles have the same observations 5 silent Suppress messages about parsing Default is false Pmetrics User s Guide 41 eC Plotting There are numerous plotting methods included in Pmetrics to generate standardized but customizable graphical visualizations of Pmetrics data Taking advantage of the class attribute in R a single plot command is used to access all of the appropriate plot methods for each Pmetrics object class To access the R help for these methods you must query each method specifically to get details for plot will only give you the parent function PMop PMload plot PMop Plot population or individual Bayesian makeOP posterior predicted data vs observed Optionally you can generate residual plots PMfinal PMload plot PMfinal Plot marginal final cycle parameter makeFinal value distributions Specifying a formula of the form y x will generate a bivariate plot For NPAG only a formula of the form prob x will plot the Bayesian posterior parameter x distributions for included subjects PMcycle PMload plot PMcycle Plots a panel with the following makeCycle windows 2 times the log likelihood at each cycle gamma lambda at each cycle Akaike Information Criterion at each cyle and Bayesian Schwartz Information Criterion at each cycle the mean parameter
62. number of the simulation id is the simulated profile number within each simulation target is the specified target and pdi is the target pharmacodynamic index e g time gt target auc target etc Outcome A data frame summarizing the results with the following columns simnum and target are as for results prop success column has the proportion with a pdi success as specified in the function call pdi mean and pdi sd columns have the mean and standard deviation of the target pharmacodynamic index e g proportion end start above target ratio of Cmax to target for each simulation and target If targets was specified via makePTAtarget to be a sampled distribution then the target column will be missing from the outcome table PMpta objects can be summarized with summary x and plotted with plot x makePop These functions create data frames of class PMpop and makePop makePost PMpost respectively makePost The PMpop or PMpost object contains several columns NDAN oniy as described in the previous section Pmetrics User s Guide 26 commana makeNPDE This function is a Pmetrics wrapper to the autoNPDE makeNPDE function in the NPDE package of Comets et al NPDE autoNPDE automatically loaded with Pmetrics that will generate NPDE NpdeObject an PMnpde object which is a list of NpdeObjects one for each output equation NpdeObjects contain normalized prediction distribution errors For a given NPAG or IT
63. of columns equal to the number of parameters npar The covariance matrix will be divided by length weights and applied to each distribution Examples SIMrun poppar final 1 SIMrun poppar list weight 1 mean c 2 4 covar diag 2 2 SIMrun poppar list weight c 0 75 0 25 mean matrix c 2 4 1 2 nrow 2 byrow T covar diag 2 2 The first example will use the population parameter distribution from Run 1 NPAG or IT2B The second example will use a distribution for two parameters with mean values of 2 and 4 and a covariance matrix with variances of 2 the diagonals and covariances of 0 the off diagonals The third example will simulate from a bimodal bivariate distribution The first distribution with weight 0 75 has means of 2 and 4 for the parameters The second distribution with weight 0 25 has means of 1 and 2 The overall covariance is the same as in example 2 but will be divided evenly across the two distributions Example 3 is similar to what occurs with a PMfinal poppar object when split 5 below is true 2 limits If limits are specified each simulated parameter set that contains any parameter value outside of the limits will be ignored and another set will be generated This will result in a truncated distribution See SIMparse for details on the full and truncated distributions returned by the simulator Four options exist for limits 1 The default NULL indicates that no limits are to be ap
64. olumn will be missing from the outcome table e prop success column has the proportion with a pdi gt success as specified in the function call e The pdi mean and pdi sd columns have the mean and standard deviation of the target pharmacodynamic index e g proportion end start above target ratio of Cmax to target for each simulation and target PMpta objects can be summarized with summary x and plotted with plot x See summary PMpta and plot PMpta in R for additional help Model Diagnostics Internal Validation Several tools are available in Pmetrics to assist with model selection The simplest methods are using PMcompare and plot PMop via the plot command for a PMop object made by makeOP or by using PMload after a successful run PMstep is another option for covariate analysis All these functions are carefully documented within R and accessible using the command or help command syntax Plotting observed vs predicted objects class PMop When you have successfully completed an NPAG or IT2B run and loaded the results with PMload one of the objects will be a PMop object Example PMload 1 will include op 1 in the list of loaded objects You can also remake the PMop object for example if you want use the mean of parameter values for calculation of predictions rather than the default median op 1 lt makeOP NPdata 1 mean In both cases you can then simply plot op 1 which is a wrapper for plot PMop
65. onding to the run number within the parentheses of the loading functions e g PM1oad 1 which will suffix all the above objects with that integer e g op 1 final 1 NPdata 1 This allows several models to be loaded into R simultaneously each with a unique suffix and which can be compared with the PMcompare command see Model Diagnostics below Pmetrics User s Guide 23 Making New Pmetrics Objects Once you have loaded the raw NPdata or ITdata and mdata or processed op final cycle pop post data objects described above with PMload run num should you wish to remake the processed objects with parameters other than the defaults you can easily do so with the make family of commands For example the default for PMop observed vs predicted objects is to use the prediction based on the median of the population or posterior distribution If you wish to use the mean of the distribution remake the PMop object using makeOP If you wish to see all the cycle information in a PMcycle object not omitting the first 1096 of cycles by default remake it using makeCycle For all of the following commands the data input is either NPdata or ITdata with additional function arguments specific to each command Accessing the help for each function in R will provide further details on the arguments defaults and output of each command makeAUC Make a data frame of class PMauc containing subject ID nakeAUC and AUC from a variety of inputs incl
66. op 1 See plot PMop for options to this plot One particular option to plot PMop resid T will generate a residual plot instead of an observed vs predicted plot A residual plot consists of three panels 1 weighted residuals predicted observed vs time 2 Pmetrics User s Guide 49 eC weighted residuals vs predictions 3 a histogram of residuals with a superimposed normal curve if the option ref T is specified the default The mean of the weighted residuals expected to be 0 is reported along with the probability that it is different from 0 by chance Three tests of normality are reported for the residuals D Agostino 2 Shapiro Wilk and Kolmogorov Smirnof An example is shown in the Plotting section Covariate analysis The PMstep function will report a 2x2 matrix of P values for the linear regression coefficients of each subject s covariates vs Bayesian posterior parameter values from the PMcov object loaded with PMload Example PMstep cov 1 Entries in the matrix are the multivariate P values A value of NA indicates that the variable was not retained in the final model The default method is both but forward selection and backward elimination are possible PMstep uses the step function in the stats package for R which is a default package It is possible to test non linear relationships using capabilities of R and the PMcov object for example with the n1s function for non linear least squares analysis Mo
67. ou wish to have installed Description Description ip Components GNU Make for MinGW 8 gcc TDM64 Current 4 6 1 tdm64 1 NOT for MSYS Version v binutils Binutls CVS 2 21 53 20110731 tdm6 TDM64 Current 4 6 1 tdm64 1 2 V mingw64 runtime MinGW w64 Runtime Snaps vj core v mingw32 make MinGW Stable 3 82 5 Iv c xv fe gdb Stable Release 7 3 1 tdm64 1 Cerra E m r 3 4 m Download 31 2 MB Install 260 MB Download 36 7 MB Install 296 MB a on Linux users have the easiest time as gfortran comes with Linux Rstudio A text editor that can link to R is useful for saving scripts Both the Windows and Mac versions of R have rudimentary text editors that are stable and reliable Numerous other free and paid editors can also do the job and these can be located by searching the internet We prefer Rstudio ul ilu i Pmetrics User s Guide 4 What This Manual Is Not We assume that the user has familiarity with population modeling and R and thus this manual is not a tutorial for basic concepts and techniques in either domain We have tried to make the R code simple regular and well documented A very good free online resource for learning the basics of R can be found at http www statmethods net index html We recognize that initial use of a new software package can be complex so please
68. ours if the number of intervals is not greater than 48 otherwise it defaults to the interval which allows for 48 intervals Examples NPrun aucint 12 This example will use an interval of 12 hours for AUC estimations directly by NPAG Note that the command makeAUC can be used to calculate AUCs of any interval from the output idelta Interval in minutes for predictions at times other than observations Default is 12 minutes which in general for most models provides sufficient granularity Smaller values can result in very large files for big populations Examples NPrun idelta 60 This generates predictions every 60 minutes prior Name ofa suitable NPAG output object from a prior run loaded with PMload i e the NPdata object A prior may be specified if the user wishes to start from a non uniform prior distribution for the NPAG run The default value is 99 which translates in NPAG to a uniform prior distribution An alternative is to include a DENO0001 file from the prior NPAG run in the working directory of the new run and specify this as the value for prior e g prior DEN0001 Examples NPrun data 2 model 2 prior NPdata 2 This example will use the model data and population density files from Run 2 to continue for another 100 cycles This is a useful way to continue a previous run that has not converged See also cycles 9 above for a discussion and example of specifying a prior simply to calculate the Bayesian posterior paramete
69. pecified by outname otherwise the filename will be simout In either case integers 1 to nsub will be appended to outname or simout e g simout1 txt simout2 txt Output files from the simulator can be read into R using the SIMparse command see documentation in R There is a plot method plot PMsim for objects created by SIMparse Here we detail the arguments to the SIMrun function SIMrun poppar limits NULL model model txt data data csv split F include exclude nsim 1000 predInt 0 seed 17 ode 4 obsNoise doseTimeNoise rep 0 4 doseNoise rep 0 4 obsTimeNoise rep 0 4 makecsv outname cleanUp T silent F 1 poppar Either an object of class PMfinal or a list containing three items in this order but of any name vector of weights vector of mean parameter values and a covariance matrix By far the easiest is to use the PMfinal object However the list option is available for example to simulate with values obtained from the literature or other sources For the list option if only one distribution is to be specified the weights vector should be of length 1 and contain a 1 If multiple distributions are to be sampled the weights vector should be of length equal to the number of distributions and its values should sum to 1 e g c 0 25 0 05 0 7 The means matrix may be a vector for a single distribution or a matrix with length weights rows and number
70. ples ITrun xdev 3 Recommend parameter ranges which are equal to the mean 3 SD You might choose a smaller range like this if you have log transformed your parameters 11 icen Same as for NPAG 12 overwrite Same as for NPAG 13 nocheck Same as for NPAG The IT2B run can complete in seconds for small populations with analytic solutions or days for large populations with complex differential equations At the end of a successful run the results will be automatically parsed and saved to the output directory Your default browser will launch with a summary of the run ERR Runs Here we provide details of the arguments available to the ERRrun command This function is identical to ITrun but the purpose is to estimate the assay error coefficients CO C1 C2 C3 from the data if you have no idea what they are Note that the estimates you get will be dependent on the model You must have a data and Pmetrics User s Guide 35 eC model file in your working directory In the R syntax below any argument with a value has a default equal to that value and if you wish to use that default you do not have to specify the argument in your function call to ERRrun ERRrun model model txt data data csv run include exclude ode 4 salt cycles 100 search cursory tol 0 001 xdev 5 overwrite F model Same as for NPAG IT2B data Same as for NPAG IT2B run Same as for NPAG IT2B include Same a
71. plied to simulated parameters 2 The second option is to set limits to NA This will use the parameter limits on the primary parameters that are specified in the model file 3 The third option is a numeric vector of length 1 or 2 e g 3 or c 0 5 4 which specifies what to multiply the columns of the limits in the model file If length 1 then the lower limits will be the same as in the model file and the upper limits will be multiplied by value specified If length 2 then the lower and upper Pmetrics User s Guide 37 Kee limits will be multiplied by the specified values If this option is used popppar must be a PMfinal object 4 The fourth option for limits is a fully customized matrix of limits for simulated values for each parameter which will overwrite any limits in the model file If specified it should be a data frame or matrix with number of rows equal to the number of random parameters and 2 columns corresponding to the minimum and maximum values For example a final ab object or a directly coded matrix e g matrix c 0 5 0 5 0 01 100 nrow 3 ncol 2 byrow T for 3 parameters with limits of 0 5 0 5 and 0 01 100 respectively It is possible to convert a parameter to fixed by omitting the second limit Means and covariances of the total number of simulated sets will be returned to verify the simulation but only those sets within the specified limits will be used to generate output s and the means and covariances of the retained s
72. r and output values for a population overwrite Overwrite existing run result folders Only relevant when the run 3 argument is specified Default is FALSE Examples NPrun run 2 overwrite T This example will delete the previous Run 2 and replace it with the current run nocheck The default is false Set to true to suppress automatic checking of data files for errors with PMcheck which depends on Java If Java is out of date or is not a 64 bit version on 64 bit systems PMcheck can fail Update Java ensure 64 bit if applicable but use this argument as a temporary fix parallel As of version 1 4 0 Pmetrics can run in parallel processing mode The default is NA which allows Pmetrics to choose parallel mode for models with differential equations or serial mode for algebraic models This is because the overhead of managing the parallel threads is greater than the efficiency of algebraic processing NOTE Parallel processing is not possible on 32 bit systems and this argument will always be set to false Pmetrics User s Guide 34 eC The NPAG run can complete in seconds for small populations with analytic solutions or days for large populations with complex differential equations At the end of a successful run the results will be automatically parsed and saved to the output directory Your default browser will launch with a summary of the run IT2B Runs Here we provide details of the arguments available to the I Trun comma
73. r conditional relationships Example Cov wt cyp IC where IC will be piece wise constant and the other two will be linearly interpolated for missing values Secondary variables Secondary variables are those that are defined by equations that are combinations of primary covariates and other secondary variables If using other secondary variables define them first within this block Equation syntax must be Fortran It is permissible to have conditional statements but because expressions in this block are Pmetrics User s Guide 11 Kee translated into variable declarations in Fortran expressions other than of the form X function Y must be prefixed by a and contain only variables which have been previously defined in the Primary Covariate or Secondary blocks Example Sec CL Ke V wt 0 75 IF cyp GT 1 CL CL cyp Bolus inputs By default inputs with DUR duration of 0 in the data csv file are delivered instantaneously to the model compartment equal to the input number i e input 1 goes to compartment 1 input 2 goes to compartment 2 etc This can be overridden with NBOLUS input number compartment number Example Bol NBCOMP 1 2 Initial conditions By default all model compartments have zero amounts at time 0 This can be changed by specifying the compartment amount as X expression where is the compartment number Primary and secondary variables and covariates ma
74. rom the data Finally you can use a generic set of coefficients We recommend that as a start Co be set to half of the lowest concentration in the dataset and C be set to 0 15 C2 and C3 can be 0 In the multiplicative model gamma is a scalar on SD In general well designed and executed studies will have data with gamma values approaching 1 Poor quality noisy data will result in gammas of 5 or more Lambda is an additive model to capture process noise rather than the multiplicative gamma model Pmetrics User s Guide 13 eC To specify the model in this block the first line needs to be either L number or G number for a lambda or gamma error model The number term is the starting value for lambda or gamma Good starting values for lambda are 1 times Co for good quality data 3 times Co for medium and 5 or 10 times Co for poor quality Note that Co should generally not be 0 as it represents machine noise e g HPLC or mass spectrometer that is always present For gamma good starting values are 1 for high quality data 3 for medium and 5 or 10 for poor quality If you include an exclamation point in the declaration then lambda or gamma will be fixed and not estimated Note that you can only fix lambda currently to zero The next line s contain the values for Co Ci C2 and C3 separated by commas There should be one line of coefficients for each output equation By default Pmetrics will use values for these coefficients
75. rors when trying to execute this command Parameter value distributions in linear space run the risk of simulating extreme or even inappropriately negative parameter values which can in turn lead to simulated observations far beyond anything corresponding to possible reality Arguments to the Pmetrics simulator can be supplied in the makeNPDE call such as split and limits to mitigate this problem In Pmetrics makeNPDE automatically generates the simulations necessary for a VPC VPCs are cumbersome when models include covariates or have heterogeneous dosing sampling regimens among subjects in the population Itis nonetheless possible to obtain a VPC in two ways The first way to generate a VPC is to use the PMsim object saved after running ma keNPDE which will be called sim x where x is the run number If an observed vs predicted PMop object made with makeOP is passed to plot PMsim with the obs argument the observed values will be overlaid upon simulated profiles if possible and an NPC will be returned in addition to the plot The NPC is simply a binomial test for the percentage of observations less than the quantiles specified by the probs argument 0 05 0 25 0 5 0 75 0 95 by default It is up to the user to decide if the study population and model is homogeneous enough to justify a VPC Example vpc plot sim 1 obs op 1 Note that when we use the obs argument the plot call will return a list of in addition to a p
76. s for NPAG IT2B exclude Same as for NPAG IT2B ode Same as for NPAG IT2B salt Same as for NPAG IT2B cycles Maximum number of ERR cycles to run Default is 100 but the value can be 1 or greater Note that unlike NPAG you cannot enter a value of 0 If you enter an integer greater than 0 the engine will terminate at convergence or the number of cycles you specify whichever comes first Examples ERRrun cycles 1000 The first example will allow ERR to run 1000 cycles before terminating ON gv UI S e TD B 9 search Depth of the search for the coefficients Default is cursory but can be medium or extensive which take progressively longer times to converge but are more accurate 10 tol Same as for IT2B 11 xdev Same as for IT2B 12 overwrite Same as for NPAG The ERR run can complete in seconds for small populations with analytic solutions or days for large populations with complex differential equations At the end of a successful run the results will be automatically parsed and saved to the output directory Your default browser will launch with a summary of the run that includes the estimates of the assay error coefficients for each output equation in the model Simulator Runs The Pmetrics simulator is a powerful Monte Carlo engine that is smoothly integrated with Pmetrics inputs and outputs Unlike NPAG and IT2B it is run from within R No batch file is created or terminal window opened However the actual si
77. s with PMload run num2 and plot etc as usual Pmetrics Outputs At the end of a successful NPAG IT2B or ERR run your Run folder will have a new folder with the run number which is sequentially designated from the previous run Within this folder are 4 subfolders Pmetrics User s Guide 52 eC e etc This folder contains files made during the run Generally you do not need to look at these They include the fortran code specifying the model model for and the instructions instr inx that NPAG or IT2B used e inputs This folder contains the original data and model file used in the run e outputs This folder contains the NPAGout Rdata or IT2B Rdata file see Data Objects which are created by the PMreport function as well as the NPAGreport html or IT2Breport html files which you will see open automatically in your default browser at the end of a run This html generates a tabbed page which will display properly as long as it is in the run outputs folder where the supporting image files are also located Within this folder are also csv comma separated values tables of the population parameter value summaries popparam csv covariance matrix popcov csv correlation matrix popcor csv and points NPAG only poppoints csv Images for default graphs are available as png and pdf files Additional text files for NPAG OUTO001 OUTTO001 DENO0001 ILOGOO01 PRTB0001 NP RF0001 txt contain combined cycle output density cycle outp
78. se in mg kg per subject summ lt summary mdata 1 formula I dose wt id FUN mean summ formula id I dose wt Pmetrics User s Guide 30 12 847966 2 9 022556 2 12 847966 4 11 811024 5 9 118541 6 9 230769 7 11 605416 8 11 718175 B 10 908097 10 121 59 0156315 11 10 619469 12 12 526096 13 9 917255 14 10 135135 15 10 465116 16 9 31677 Jy 10 948905 18 10 158014 19 12 20 10 169492 mean s 1 10 84424 PMop summary x median summary PMop digits outeq umm formula 2 max 3 choose the second column getOption digits This will summarize a PMop object that contains observations predictions and errors You can specify the number of digits predictions based on posterior or population predictions or based on mean or median parameter distributions for any output equation Example data PMex1 summary op 1 Time Obs disummarize based on medians of Pred Pmetrics User s Guide 31 Bayesian posterior parameter values Min 120 0000 1 8600 1 9648 259 121 0250 4 4950 4 5019 Median 126 0000 6 5600 6 5996 715 132 0000 28 898850 8 9355 Max 144 9800 20 1500 19 9025 Mean 127 6899 7 2407 7 0689 SD 7 8000 3 7992 3 0420 Mean prediction error 0 1718 Mean weighted prediction error bias 0 103 P 0 9718 different than 0 Mean squared prediction error 0 795 Root m
79. see Simulator Runs Example Pri KE 0 5 In this example KE has a range of 0 to 5 which V 0 01 100 will be absolute for NPAG and the simulator if KA 0 5 limits NA but defines the prior distribution KCP 5 for KE if using IT2B The limits KE to the KPC 0 5 positive real numbers for IT2B KCP is fixed to Tlag1 0 2 5 regardless of the engine IC3 0 10000 FA1 0 1 Covariates Covariates are subject specific data such as body weight contained in the data csv file The covariate names which are the column names in the data file can be included here for use in secondary variable equations The order should be the same as in the data file and although the names do not have to be the same we strongly encourage you to make them the same to avoid confusion Covariates are applied at each dose event The first dose event for each subject must have a value for every covariate in the data file By default missing covariate values for subsequent dose events are linearly interpolated between existing values or carried forward if the first value is the only non missing entry To suppress interpolation and carry forward the previous value in a piece wise constant fashion include an exclamation point 1 in any declaration line Note that any covariate relationship to any parameter may be described as the user wishes by mathematical equations and Fortran code allowing for exploration of complex non linear time dependent and o
80. simout or a list of PMsim objects made by SIMparse with suitable simulated regimens and observations e g simdata lt SIMparse simout txt The latter method is preferable so that you do not have to re parse the simulator output each time you run makePTA The number and times of simulated observations does not have to be the same in all objects 2 simlabels Optional character vector of labels for each simulation Default is c Regimen 1 Regimen 2 These will be used in plots 3 targets A vector of pharmacodynamic targets such as Minimum Inhibitory Concentrations MICs This can also be a sampled distribution using makePTAtarget which generates an object of class PMpta targ that can be used in makePTA The single input to makePTAtarget is a data frame or name of csv file in working directory whose first two columns are targets and the number of samples for each target An example can be seen for Staphylococcus aureus susceptibility to vancomycin at the EUCAST website at http mic eucast org Eucast2 regShow jsp Id 1214 see the bottom of the page Examples Pmetrics User s Guide 47 eC makePTA targets c 0 25 0 5 1 2 4 8 16 32 will test all profiles against the specific targets listed On the other hand makePTA targets makePTAtarget staph csv will test each profile against one MIC drawn from a distribution of MICs for contained in the staph csv file 4 target type A numeric
81. sing Example simdata SIMparse simout txt include c 1 4 If the wildcard match returned four files simouti txt simout2 txt simout3 txt and simout4 txt and you wished to only parse the second and third files this would accomplish that filtering Be careful using include and exclude at the same time you might end up with no files 4 combine Boolean parameter default False which specifies whether you wish to combine the parsed files into a single PMsim object This can be useful for making visual predictive checks for example If combine F and multiple files are parsed then the return object will be a list of PMsim objects which can be plotted or otherwise accessed using standard list referencing eg plot simlist 1 plot simlist 2 etc In this case each element of the list corresponds to a single simulation template Examples simdata lt SIMparse simout txt combine F simdata lt SIMparse simout txt combine F If your template data file had four subjects and you did not combine the results as in the first example simdata would be a list of length 4 simdata 1 simdata 2 simdata 31 simdata 4 each containing a PMsim simulation output object corresponding to a subject in the template data file In the second example all 4 simulation outputs would be combined into a single PMsim object You can combine any simulator output files even with differing numbers of simulated parameter sets The
82. so prior 14 below indpts Index of starting grid point number Default is missing which allows NPAG to choose depending on the number of random parameters 1 or 2 parameters index of 1 3 3 4 4 5 6 6 or more is 10 number of multiples for each parameter greater than 5 e g 6 101 7 102 up to 108 for 13 or more parameters This number corresponds to the number of grid support points which will initially fill the model parameter space The larger the number of random parameters to be estimated the more points are required The more you choose the slower the run will be but results may improve It is reasonable to choose fewer points early in model exploration and increase in later phases or if poor model fits or lack of convergence are noted Indices of 1 to 6 correspond respectively to 2129 5003 10007 20011 40009 and 80021 points 101 will choose 80021 points plus one additional multiple of 80021 points Examples NPrun indpts 102 This will allow NPAG to start with 80021 2 80021 240063 points The unusual numbers are from early development of NPAG where primes were thought to be important icen Summary of parameter distributions to be used to calculate predictions Default is median and other choices are mean or mode Examples NPrun icen mean This example will use the means of the parameter distributions rather than the medians to calculate predictions aucint Interval for AUC calculations Default is 24 h
83. st Data csv file Model txt file Preparation program Instruction file Engine program Output R is used to specify the working directory containing the data csv and model txt files Through the batch file generated by R the preparation program is compiled and executed The instruction file is generated automatically by the contents of the data and model files and by arguments to the NPrun ITrun or ERRrun commands The batch file will then compile and execute the engine file according to the instructions which will generate several output files upon completion Finally the batch file will call the R script to generate the summary report and several data objects including the IT2Bout Rdata or NPAGout Rdata files which can be loaded into R subsequently using PMload Objects that are modified can be saved back to the Rdata files with PMsave Both input files data model are text files which can be edited directly Pmetrics User s Guide 7 Pmetrics Input Files Data csv Files Pmetrics accepts input as a spreadsheet matrix format It is designed for input of multiple records in a concise way Please keep the number of characters in the file name s 8 Files are in comma separated values csv format Examples of programs that can save csv files are any text editor e g TextEdit on Mac Notepad on Windows or spreadsheet program e g Excel Click on hyperlinked items to see an explanation IMPORT
84. th 4 e g 40 1 D 1 0 0X 16 obsTimeNoise A vector of observation timing error polynomial coefficients The default is 0 for all coefficients This is applied to all observation times regardless of output number so can only be of length 4 e g amp 0 1 0 1 0 0 17 makecsv A character vector for the name of the single csv file to be made for all simulated subjects If missing no files will be made Examples SIMrun makecsv newsubj csv This example will make a csv file in Pmetrics format called newsubj csv containing the simulated subjects If there were 3 subjects in the template data file and nsim 10 then newsubj csv will contain 3 10 30 simulated ubject records Pmetrics User s Guide 40 eC 18 outname The name for the simulator output file s without an extension Numbers 1 to nsub will be appended to the files If missing will default to simout Examples SIMrun outname sim2 This will make the following output files for a template with 3 subjects sim2 1 txt sim2 2 txt sim2 3 txt These are the files that are parsed by SIMparse 19 clean This is a rarely used Boolean parameter to specify whether temporary files made in the course of the simulation run should be deleted It is primarily used for debugging Defaults to True 20 silent Boolean operator controlling whether a model summary report is given Default is False To get the results of a simulator run back into R you need to use
85. to 0 9 9 cycles Maximum number of NPAG cycles to run Default is 100 but the value can be 0 or greater If you enter an integer greater than 0 the engine will terminate at convergence or the number of cycles you specify whichever comes first Early in model exploration values of 10 to 100 can be useful with larger values later in model development In order to facilitate model comparison however we recommend using the same cycle limit for all early models e g 100 rather than choosing 10 for one and 100 for another If you enter 0 this is the way to test the predictive power of a model on an independent data set and a non uniform prior must be Pmetrics User s Guide 33 eC 10 11 12 13 14 15 16 17 specified This means that the engine will only calculate the individual Bayesian posteriors for the new subjects using the population joint density from a previous run as a Bayesian prior Examples NPrun cycles 1000 NPrun data newdata csv cycles 0 prior NPdata 1 The first example will allow NPAG to run 1000 cycles before terminating The second example calculates Bayesian posteriors only i e NPAG is not cycling using the default model file and some new data Note by specifying a prior which is required with 0 cycles the grid points are distributed in a non uniform manner corresponding to the prior distribution in NPdata 1 the output of Run 1 This means that the indpts argument will be ignored See al
86. ty weight of that set of estimates There can be at most one point for each subject in the study population There is no need for any assumption about the underlying distribution of model parameter values The simulator is a semi parametric Monte Carlo simulation software program that can use the output of IT2B or NPAG to build randomly generated response profiles e g time concentration curves for a given population model parameter estimates and data input Simulation from a non parametric joint density model i e NPAG output is possible with each point serving as the mean of a multivariate normal distribution weighted according to the weight of the point The covariance matrix of the entire set of support points is divided equally among the points for the purposes of simulation Pmetrics User s Guide 5 eC Pmetrics has groups of R functions named logically to run each of these programs and to extract the output Again these are extensively documented within R by using the help command or command syntax JTrun ITparse ERRrun NPrun NPparse e PMload PMsave PMreport SIMrun SIMparse For IT2B and NPAG the run functions generate batch files which when executed launch the software programs to do the analysis ERRrun is a special implementation of IT2B designed to estimate the assay error polynomial coefficients from the data when they cannot be calculated from assay validation data using makeErrorPoly supplied
87. uding objects of PMop PMsim PMpop PMpost or a suitable data frame makeCov Generate a data frame of class PMcov with subject makeCov specific covariates extracted from the data csv file This object can be plotted and used to test for covariates which are significantly associated with model parameters makeCycle Create a PMcycle object described in the previous makeCycle section Create a PMfinal object described in the previous section Create a PMop object described in the previous section Pmetrics User s Guide 24 makeNCA Create a data frame class PMnca with the outputofa makeNCA non compartmental analysis using PMmatrix or PMpost data objects as input The PMnca object contains several columns id Subject identification auc Area under the time observation curve using the trapezoidal approximation from time 0 until the second dose or if only one dose until the last observation aumc Area under the first moment curve k Slope by least squares linear regression of the final 6 log transformed observations vs time auclast Area under the curve from the time of the last observation to infinity calculated as Final obs k aumclast Area under the first moment curve from the time of the last observation to infinity aucinf Area under the curve from time 0 to infinity caluculated as auc auclast aumcinf Area under the first moment curve from time 0 to infinity mrt Mean residence time calculated as 1 k cm
88. ult is model txt This file will be converted to a fortran model file If it is detected to already be a fortran file then the analysis will proceed without any further file conversion Examples NPrun NPrun model model2 txt NPrun model 2 The first example uses all the default options The second example uses the defaults except for the model file The third example uses the same model as that used in Run 2 but defaults otherwise 2 data Name ofa suitable data file or an existing previous run number corresponding to a folder in the current working directory that used the same data file as will be used in the current run If this is supplied then previously made ZMQ files will be copied into the current working directory bypassing the need to re convert the csv file and speeding up the run Examples NPrun data data2 csv NPrun data 2 The first example uses the defaults except for the data file The second example uses the same data file as that used in Run 2 but defaults otherwise 3 run Specify the run number of the output folder Default if missing is the next available number Examples NPrun NPrun run 2 The first example uses all the default options and the run number will be automatically assigned to 3 if runs 1 and 2 already exist The second example uses the defaults except for the run number If runs 1 and 3 exist and you have deleted run 2 because you wish to re run it then this syntax will place the output i
89. un 1 Ka Kel V all subjects setwd working directory NPrun assumes model model txt and data data csv PMload 1 Remember in R that the command example function will provide examples for the specified function Most Pmetrics functions have examples Pmetrics Data Objects After a successful IT2B or NPAG run an R datafile is saved in the output subdirectory of the newly created numerically ordered folder in the working directory After IT2B this file is called IT2Bout Rdata and after NPAG it is called NPAGout Rdata As mentioned in the previous section these data files can be loaded by ensuring that the Runs folder is set as the working directory and then using the Pmetrics commands PMload run num There are several Pmetrics data objects contained within the Rdata files which are loaded with PM1oad making these objects available for plotting and other analysis Objects loaded by PMload run num E eme emen deme e mmm E eme mmm 00 pred type Type of prediction i e based on the population parameter values or Bayesian posterior parameter values icen Median default or mean of the parameter distributions used to calculate the predicted values BEEN KNIE EN Pmetrics User s Guide 19 block Dosing block usually 1 unless data file contains EVID 4 dose reset events in which case each such reset within a given ID will increment the dosing block by 1 for that ID obsSD Calculated stand
90. ut density convergence posterior predictions and R friendly information respectively For IT2B these text files OUTF0001 OUFF0001 DENF0001 LASTO001 FROMO0001 IT RF0001 txt contain combined cycle output density cycle output density Bayesian posterior parameters parameter ranges that can be passed to NPAG and R friendly information respectively Normally you do not need to look at these files Finally for outputs there is an NPAGreport tex or IT2Breport tex file For users familiar with LATEX this file can be used to generate a pdf version of the html report page useful for sharing with colleagues You must have an installed LATEXengine Our recommended LATEX engine for Mac users is MacTex which is very large gt 1GB but contains a complete installation For Windows MikTex is our recommended engine Both are free Once you have a IATEX engine it is easy to make a pdf from the tex file Simply open it with Rstudio and click the Make PDF button that appears The pdf file will be in your outputs folder e wrkcopy This folder contains working copy files in the original format used by NPAG and IT2B and which are still used under the hood They are easy to read and can be used as a check if something goes wrong Ancillary Functions Pmetrics has some ancillary functions which can be helpful when you are involved in population PK modeling These functions do not control the components of Pmetrics or process output
91. uts are of the form Y expression where is the output equation number Primary and secondary variables and covariates may be used in the expression as can conditional statemtents in Fortran code A prefix is not necessary in this block for any statement although if present will be ignored There can be a maximum of 6 outputs They are referred to as Y 1 Y 2 etc Example Out Y 1 X 2 V Error This block contains all the information Pmetrics requires for the structure of the error model In Pmetrics each observation is weighted by 1 error There are two choices for the error term 1 error SD gamma 2 error SD lamda Note that lambda is only available in NPAG currently where SD is the standard deviation SD of each observation obs and gamma and lambda are terms to capture extra process noise related to the observation including mis specified dosing and observation times SD is modeled by a polynomial equation with up to four terms Co Ci obs C2 obs Ca obs The values for the coefficients should ideally come from the analytic lab in the form of inter run standard deviations or coefficients of variation at standard concentrations You can use the Pmetrics function makeErrorPoly to choose the best set of coefficients that fit the data from the laboratory Alternatively if you have no information about the assay you can use the Pmetrics function ERRrun to estimate the coefficients f
92. y be used in the expression as can conditional statemtents in Fortran code A prefix is not necessary in this block for any statement although if present will be ignored Example Ini X 2 IC V i e IC is a covariate with the measured trough concentration prior to an observed dose X 3 IC3 i e IC3 is a fitted amount in this unobserved compartment In the first case the initial condition for compartment 2 becomes the value of the IC covariate defined in Covariate block multiplied by the current estimate of V during each iteration This is useful when a subject has been taking a drug as an outpatient and comes in to the lab for PK sampling with measurement of a concentration immediately prior to a witnessed dose which is in turn followed by more sampling In this case IC or any other covariate can be set to the initial measured concentration and if V is the volume of compartment 2 the initial condition amount in compartment 2 will now be set to the measured concentration of drug multiplied by the estimated volume for each iteration until convergence In the second case the initial condition for compartment 3 becomes another variable IC3 defined in the Primary block to fit in the model given the observed data F bioavailability nn Specify the bioavailability term if present Use the form FA expression where is the input number Primary and secondary variables and covariates may be used in the expression
93. y specified by EVID 0 event times in the data file and continuing until the maximum time in the template data file If the number of simulated observations exceeds 594 they will be truncated to this value The third example simulates observations half hourly from 1 to 12 hours plus at any EVID 0 event times in the data file covariate If you are using the results of an NPAG or IT2B run to simulate i e a PMfinal object as poppar then you can also simulate with covariates This argument is a list with names cov mean sd limits and fix e The first item is a PMcov object such as that loaded with PMload Pmetrics will use this object to calculate the correlation matrix between all covariates and Bayesian posterior parameter values e The second item mean allows you to specify a different mean for one or more of the covariates This argument is a named list where each item in the list is the name of a covariate in your data that is to have a different mean If this argument is missing then the mean covariate values in the population will be used for simulation The same applies to any covariates that are not named in the mean list e The third item in the covariate argument list is sd and this functions just as the mean object does allowing you to specify different standard deviations for covariates in the simulation If sd is missing then the standard deviations of the covariates in the population are used e The

Download Pdf Manuals

Related Search

Related Contents

Zanussi ZOU 363 User's Manual  アロマディフューザー LM-282    Stage,mode d`emploi  Anti-Alias Multichannel Module  Datasheet COMfortel® 600  MANUAL DE INSTALACIÓN  grando evo (41035413.A)  USER MANUAL Reply  ES521 - Proyectores OK  

Copyright © All rights reserved. Failed to retrieve file