Dear all, Fistly, I would very much appreciate any answer or brainstorm you
would like to give on this topic. The core of my question is whether SPE is
still an ideal metric for model evaluation? Is it equal to WRES?
I have recently encountered a very interesting metic called "SPE", standardized
prediction error, derived from "J Pharmacokinetic Biopharm 1990 18(2) 161-73
Evaluation of PPK Parameter Estimates". The calculation method of SPE is
almost the same as that of WRES. So are these 2 just synonyms? Besides, what's
the relation between SPE，WRES and CWRES？ After observing the output of the
control file,
$ESTIMATION MAXEVAL=0 METHOD=1 INTE NOABORT POSTHOC $TABLE ID TIME
IPRED MDV...CWRES NOPRINT ONEHEAD FILE=1.fit I found the results of CWRES are
totally different from WRES. Since we used FOCEI for estimation, is CWRES
preferable and we can just ignore WRES results? Finally, what are the best
metrics for external valuation? Prediction Error(=(Pedicted Conc.- Observed
Conc)/Observed Conc); CWRES; PD (Prediction discrepancies) or NPDE?Can one be
substitued totally by another? THX All In AdvanceZhao Chenyan 赵晨妍Department of
Clinical Pharmacy, Huashan Hospital, Fudan UniversityNo.12 Middle Urumqi Road,
Jing‘an District, Shanghai, 200040, P.R.ChinaTEL：+86 13917430219
