Re: NPDE Calculation

2013/11/22 <[email protected]> > Tao, > > > > Yes the covariance matrix is computed for each subject separately (all the > observations of one subject). > > > > Best regards, > > > > Karl > > > > *De :* Tao Liu [mailto:[email protected]] > *Envoyé :* vendredi 22 novembre 2013 16:24 > *À :* BRENDEL Karl IRIS > *Cc :* [email protected] > *Objet :* Re: [NMusers] NPDE Calculation > > > > Karl-Thanks for your reply. As you said 'Cholesky decomposition is used > to obtain the inverse square root of the covariance matrix', so my question > is the covariance matrix of what? Let's say I have 5 subjects and 10 > observations per subject. The covariance matrix is for each subject > separately? (10*10 cov matrix), or each observation separately? (1*1 cov > matrix), or for all the observations? (50*50 cov matrix) > > > > In France Mentr´ (2006) Equation 9 speij = (yij − E(yi j ))/SD(yi j). This > means spe should be calculated for each observation separately? (1*1 cov > matrix ?, actually, there's no covariance here) > > > > In E. Comets (2008), Yi, sim(k) was defined as the vector of simulated > observations for the ith subject in the kth simulation and E(Yi) is the > expectation (mean) of Yi based on all the simulated Yi , sim(k) . And all > the following computation is based on Yi, a vector. So, it means 10*10 > cov matrix here? > > > > Thanks, > > > > Tao > > > > 2013/11/22 <[email protected]> > > Dear Tao, > > > > Cholesky decomposition is used to obtain the inverse square root of the > covariance matrix in order to go from prediction discrepancies (pd) (1) to > prediction distribution error (PDE). Normalized prediction distribution > errors (NPDE) are obtained using the inverse function of the normal > cumulative function (2). > > So if you have more than one observation per subject, it is better to > compute NPDE. > > > > To ways to compute easily NPDE: > > · use the R package an read the paper from Emmanuelle Comets(3) (the > numerical results are saved in a file with extension .npde (the name of > which is given by the user; the file contains the components id, xobs, > ypred, npde) > > > > · NPDE are also available in NONMEM7 > > > > $TAB > > NPDE CWRES DV PRED RES WRES > > > > ESAMPLE=1000 SEED=1234567; number of simulation used to compute NPDE and > the seed > > ONEHEADER NOAPPEND NOPRINT FILE=EXAMPLE.TAB > > > > Best regards, > > > > Karl > > > > (1) F. Mentré, S. Escolano, Prediction discrepancies for the evaluation > of nonlinear mixed-effects models, J.Pharmacokinet. Biopharm. 33 (2006) > 345–367. > > (2) K. Brendel, E. Comets, C. Laffont, C. Laveille, F. Mentré. Metrics > for external model evaluation with an application to the population > pharmacokinetics of gliclazide, Pharm. Res.23 (2006) 2036–2049. > > (3) E. Comets, K.Brendel, F. Mentré. Computing normalised prediction > distribution errors to evaluate nonlinear mixed-effect models:The npde > add-on package for R. > > > > > > > > *De :* [email protected] [mailto:[email protected]] *De > la part de* Tao Liu > *Envoyé :* vendredi 22 novembre 2013 03:55 > *À :* [email protected] > *Objet :* [NMusers] NPDE Calculation > > > > Dear Nmusers, > > > > I have two questions regarding to the calculation of NPDE and the > corresponding package in R. > > > > 1) In Cholesky decompsition, should I calculate pd for each observation > separately using corresponding mean and sd, or I should use the var-cov > matrix and calculate pd for all the observations simultaneously? Should I > consider the covariance between different observations here? > > > > 2) How can I save the individual NPDE value from the autonpde function in > R ? > > > > Thanks, > > > > -- > > > > Tao Liu BSc > PhD Student > > > > Center for Translational Medicine > University of Maryland, Baltimore > > http://www.ctm.umaryland.edu/ > > > Email: [email protected] > > ＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝ > The information in this email is confidential, and is intended solely for > the addressee(s). Access to this email by anyone else is unauthorized and > therefore prohibited. If you are not the intended recipient you are > notified that disclosing, copying, distributing or taking any action in > reliance on the contents of this information is strictly prohibited and may > be unlawful. If you are not an addressee, please inform the sender > immediately. Thanks a lot! > ＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝ > > > > > > -- > > > > Tao Liu BSc > PhD Student > > > > Center for Translational Medicine > University of Maryland, Baltimore > > http://www.ctm.umaryland.edu/ > > > Cell Phone: (443) 703-8499 > > > Email: [email protected] > > ＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝ > The information in this email is confidential, and is intended solely for > the addressee(s). Access to this email by anyone else is unauthorized and > therefore prohibited. If you are not the intended recipient you are > notified that disclosing, copying, distributing or taking any action in > reliance on the contents of this information is strictly prohibited and may > be unlawful. If you are not an addressee, please inform the sender > immediately. Thanks a lot! > ＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝ > -- Tao Liu BSc PhD Student Center for Translational Medicine University of Maryland, Baltimore http://www.ctm.umaryland.edu/ Cell Phone: (443) 703-8499 Email: [email protected] ＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝ The information in this email is confidential, and is intended solely for the addressee(s). Access to this email by anyone else is unauthorized and therefore prohibited. If you are not the intended recipient you are notified that disclosing, copying, distributing or taking any action in reliance on the contents of this information is strictly prohibited and may be unlawful. If you are not an addressee, please inform the sender immediately. Thanks a lot! ＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝