Re: [NMusers] PRED for BLQ-like observations

From: Denney, William S. <William.S.Denney_at_pfizer.com>
Date: Fri, 20 Nov 2015 21:51:31 +0000

Hi Bob and Dennis,

I was unaware of these values, thanks for the pointer. What is a use case =
for MDV=100? The only case I can think of is if you have a measurement t=
hat you don't believe to be accurate, but then it should be removed and/or =
set to actually missing before NONMEM.

Thanks,

Bill

On Nov 20, 2015, at 16:42, "Fisher Dennis" <fisher_at_plessthan.com<mailto:fis=
her_at_plessthan.com>> wrote:

Even better, take advantage of this (from NMHELP):

Values of MDV are:

      0 The DV data item is an observed value, i.e., DV is not miss-
           ing.

      1 The DV data item is not regarded an observed value, i.e., DV
           is missing. The DV data item is ignored. |

      100 Same as MDV=0, but this record is ignored during Estimation |
           and Covariance Steps. During other steps, MDV will changed |
           to 0. |

      101 Same as MDV=1, but this record is ignored during Estimation |
           and Covariance Steps. During other steps, MDV will changed |
           to 1. |

           Reserved variables MDVI1, MDVI2, MDVI3 can be used to over- |
           ride values of MDV>100. These variables are defined in |
           include file nonmem_reserved_general.

Dennis

Dennis Fisher MD
P < (The "P Less Than" Company)
Phone: 1-866-PLessThan (1-866-753-7784)
Fax: 1-866-PLessThan (1-866-753-7784)
www.PLessThan.com<https://urldefense.proofpoint.com/v2/url?u=http-3A__www=
.PLessThan.com&d=CwMFAg&c=UE1eNsedaKncO0Yl_u8bfw&r=4WqjVFXRfAkMXd6y3w=
iAtxtNlICJwFMiogoD6jkpUkg&m=WYu-CQioIB0i7YgHqkz6PNMt7uCea2R_jfzrL98PYfw&s=
=zc442lHm9Sn0NQGqUpk8TZgvysYTMDYaSmndj8HXFhY&e=>



On Nov 20, 2015, at 12:38 PM, Nick Holford <n.holford_at_auckland.ac.nz<mailto=
:n.holford_at_auckland.ac.nz>> wrote:

Pavel,
Did you test the run time with double the records?
I would expect that the MDV=1 records would be largely ignored in the est=
imation step and not contribute much to run time.
Nick

On 21-Nov-15 08:59, Pavel Belo wrote:
Thank you Bill,
In my case it exactly doubles the number of records... The records are dai=
ly measures and the code is running slow enough. I'll split the code into e=
stimation part and one that that is redundant, but uses a larger file and c=
reates an output. It will be something like
$EST MAXEVALS=9999 SIG=3 NOABORT PRINT=1 SORT CONSTRAIN=5
 METHOD=SAEM NBURN=0 NITER=0 POSTHOC INTERACTION
 LAPLACIAN GRD=TG(1-7):TS(8-9) CTYPE=3 CINTERVAL=10
I guess the best future way is modify something in NONMEM so there is an op=
tion to provide only PRED in the PRED column (version 7.4?).
Thanks!
Pavel
On Fri, Nov 20, 2015 at 01:06 PM, Denney, William S. wrote:

   Hi Pavel,

   The easiest way that I know is to generate your data file with one
   set of rows for estimation with M3 and another row just above or
   below with MDV=1. NONMEM will then provide PRED and IPRED in the
   rows with MDV=1.

   Thanks,

   Bill

   *From:*owner-nmusers_at_globomaxnm.com<mailto:owner-nmusers_at_globomaxnm.com>
   [mailto:owner-nmusers_at_globomaxnm.com] *On Behalf Of *Pavel Belo
   *Sent:* Friday, November 20, 2015 11:47 AM
   *To:* nmusers_at_globomaxnm.com<mailto:nmusers_at_globomaxnm.com>
   *Subject:* [NMusers] PRED for BLQ-like observations

   Hello The NONMEM Users,

   When we use M3-like approach, the outputs has PRED for non-missing
   observations and something else for BLQ (is that PRED=CUMD?). As
   in the diagnostic figures PRED for BLQs looks like noise, I remove
   them. It is not always perfect, but OK in for most frequent cases.

   When we use count data such as a scale with few possible values
   (for example, 0, 1, 2, 3, 4, 5), it makes more sense to use PHI
   function (home-made likelihood) for all observations rather than
   to treat the count as a continuous variable an apply M3-like
   approach to 1 and 5 while only (as we know, they are like LLOQ and
   ULOQ). In this case, all PRED values look like noise. A hard way
   to replace the noise with PRED value is to simulate PRED for each
   point and merge them with the DV and IPRED data. Is there an easy
   way?

   (The model runs well and better than when the count is treated as
   a continuous variable.)

   Thanks!

   Pavel


--
Nick Holford, Professor Clinical Pharmacology
Dept Pharmacology & Clinical Pharmacology, Bldg 503 Room 302A
University of Auckland,85 Park Rd,Private Bag 92019,Auckland,New Zealand
office:+64(9)923-6730 mobile:NZ+64(21)46 23 53
email: n.holford_at_auckland.ac.nz<mailto:n.holford_at_auckland.ac.nz>
http://holford.fmhs.auckland.ac.nz/<https://urldefense.proofpoint.com/v2/ur=
l?u=http-3A__holford.fmhs.auckland.ac.nz_&d=CwMFAg&c=UE1eNsedaKncO0Yl=
_u8bfw&r=4WqjVFXRfAkMXd6y3wiAtxtNlICJwFMiogoD6jkpUkg&m=WYu-CQioIB0i7YgH=
qkz6PNMt7uCea2R_jfzrL98PYfw&s=mXOHPHTRH3KFb_dSGnMz_dQtDkhhBtasaU3R5_x-Ip4=
&e=>

Holford SD, Allegaert K, Anderson BJ, Kukanich B, Sousa AB, Steinman A, Pyp=
endop, B., Mehvar, R., Giorgi, M., Holford,N.H.G. Parent-metabolite pharmac=
okinetic models - tests of assumptions and predictions. Journal of Pharmaco=
logy & Clinical Toxicology. 2014;2(2):1023-34.
Holford N. Clinical pharmacology = disease progression + drug action. Br =
J Clin Pharmacol. 2015;79(1):18-27.



Received on Fri Nov 20 2015 - 16:51:31 EST

This archive was generated by hypermail 2.3.0 : Fri Sep 27 2019 - 16:47:22 EDT