Dear friends,
I log transformed the plasma concentrations in my data and I am using this
log transformed data as my dependent variable (DV). I imputed the BQLs to
LLOQ/2, which is my case is (1/2=0.5). The natural log of 0.5 is negative
which mean that I have negative values in my DV column.
While the parameter estimates look somewhat reliable, I have negative
points plotted in the diagnostic plots such as DV vs PRED, DV and PRED vs
TIME, etc.
My questions are-
1. What is the best way of dealing with <1 DV/plasma concentration values
while transforming into natural logarithm?
2. In my situation discussed above, where I have negative DV values in my
data, has the NONMEM ignored or considered the negative numbers when
estimating the parameter values?
Thank you,
Manu
--
*With regards-*
*Manushree Bharadwaj, BVSc, PhD Candidate*
*Graduate Teaching Associate*
*Department of Physiological Sciences*
*Center for Veterinary Health Sciences*
*Oklahoma State University*
*Stillwater, OK-74074*
Log transformation of imputed BQL
one thing that you could do is multiply all the values by 1000 (i.e., change
the units)
Dennis Fisher MD
P < (The "P Less Than" Company)
Phone / Fax: 1-866-PLessThan (1-866-753-7784)
www.PLessThan.com http://www.plessthan.com/
Quoted reply history
> On Jul 27, 2016, at 11:11 AM, Bharadwaj, Manushree
> <[email protected]> wrote:
>
> Dear friends,
>
> I log transformed the plasma concentrations in my data and I am using this
> log transformed data as my dependent variable (DV). I imputed the BQLs to
> LLOQ/2, which is my case is (1/2=0.5). The natural log of 0.5 is negative
> which mean that I have negative values in my DV column.
> While the parameter estimates look somewhat reliable, I have negative points
> plotted in the diagnostic plots such as DV vs PRED, DV and PRED vs TIME, etc.
> My questions are-
> 1. What is the best way of dealing with <1 DV/plasma concentration values
> while transforming into natural logarithm?
> 2. In my situation discussed above, where I have negative DV values in my
> data, has the NONMEM ignored or considered the negative numbers when
> estimating the parameter values?
>
> Thank you,
> Manu
>
> --
> With regards-
> Manushree Bharadwaj, BVSc, PhD Candidate
> Graduate Teaching Associate
> Department of Physiological Sciences
> Center for Veterinary Health Sciences
> Oklahoma State University
> Stillwater, OK-74074
Dear Manu,
NONMEM doesn't care about negative values when analysing data in the log
domain. The sure-fire test is to do the analysis both after log-transforming in
the original units, and after multiplying by 1000 as Dennis suggests, and the
results should be identical (if you adjust your afflicted starting values
accordingly of course).
Kind regards,
Rik
Rik Schoemaker, PhD
Occams
Malandolaan 10
1187 HE Amstelveen
The Netherlands
http://www.occams.eu/
+31 20 441 6410
[email protected]<mailto:[email protected]>
[cid:[email protected]]
Quoted reply history
From: [email protected] [mailto:[email protected]] On
Behalf Of Dennis Fisher
Sent: 27 July 2016 23:26
To: Bharadwaj, Manushree <[email protected]>
Cc: [email protected]
Subject: Re: [NMusers] Log transformation of imputed BQL
one thing that you could do is multiply all the values by 1000 (i.e., change
the units)
Dennis Fisher MD
P < (The "P Less Than" Company)
Phone / Fax: 1-866-PLessThan (1-866-753-7784)
http://www.plessthan.com/
On Jul 27, 2016, at 11:11 AM, Bharadwaj, Manushree
<[email protected]<mailto:[email protected]>> wrote:
Dear friends,
I log transformed the plasma concentrations in my data and I am using this log
transformed data as my dependent variable (DV). I imputed the BQLs to LLOQ/2,
which is my case is (1/2=0.5). The natural log of 0.5 is negative which mean
that I have negative values in my DV column.
While the parameter estimates look somewhat reliable, I have negative points
plotted in the diagnostic plots such as DV vs PRED, DV and PRED vs TIME, etc.
My questions are-
1. What is the best way of dealing with <1 DV/plasma concentration values while
transforming into natural logarithm?
2. In my situation discussed above, where I have negative DV values in my data,
has the NONMEM ignored or considered the negative numbers when estimating the
parameter values?
Thank you,
Manu
--
With regards-
Manushree Bharadwaj, BVSc, PhD Candidate
Graduate Teaching Associate
Department of Physiological Sciences
Center for Veterinary Health Sciences
Oklahoma State University
Stillwater, OK-74074