Dear all,
Playing with repeated time to event models, I run into the issue that simple
diagnostics for a single time to event outcome suggest that constant hazard and
Weibull models are not very appropriate. The lognormal model seems to provide a
very nice fit; compared to a constant hazard, the hazard is suggested to be
higher in the beginning and then significantly lower at later times.
I have not seen any implementations online: does anyone know if the lognormal
survival function can be implemented in NONMEM, and/or can anyone suggest
alternative approaches? Some time-varying function to modify the hazard?
Any and all suggestions appreciated!
Kind regards,
Rik
Rik Schoemaker, PhD
Occams Coöperatie U.A.
Malandolaan 10
1187 HE Amstelveen
The Netherlands
http://www.occams.com
+31 20 441 6410
[email protected]<mailto:[email protected]>
[cid:[email protected]]
Lognormal survival in NONMEM?
Hej, see example code below.
Ina Frobel also applied an empirical hazard model (Frobel et al) to catch the pattern, see the model in DDMoRe model repository, with link to publication.
http://repository.ddmore.foundation/model/DDMODEL00000065
$PROBLEM Time to first event data
$SUBR ADVAN TOL=9
$MODEL COMP=(HAZARD)
$PK
SIGM= THETA(1)*EXP(ETA(1))
MU=THETA(2)
$DES
DEL= 1E-12
TIM=T+DEL
LNT = LOG(TIM)
X1 =(LNT-MU)/SIGM
PDF= EXP(-1/2*(X1**2))/SQRT(2*3.14159265)
DADT(1)=1/(TIM*SIGM)*PDF/(1-PHI(X1))
$ERROR
CHZ = A(1)
SURX = EXP(-CHZ)
DELX = 1E-12
TIMX=TIME+DELX
LNTX = LOG(TIMX)
X1X =(LNTX-MU)/SIGM
PDFX= EXP(-1/2*(X1X**2))/SQRT(2*3.14159265)
HAZNOW=1/(TIMX*SIGM)*PDFX/(1-PHI(X1X))
Y=SURX
IF(DV.EQ.1) Y=SURX*HAZNOW
$THETA (0,1) ;SIGMA ;1 SD of the log normal distribution
$THETA (0) ;MU ;2 Mean of the log normal distribution
$OMEGA 0 FIX ;OM1 ;1 ;Only to tell NONMEM that each ID has multiple rows
$ESTIM MAXEVAL99 METHOD=0 LIKE SIGL=9 NSIG=3 PRINT=1 MSFO=msfb1
BR, Siv
Siv http://katalog.uu.se/profile/?id=N96-5738, PhD
Researcher
Dept of Pharmaceutical Biosciences
Faculty of Pharmacy
Uppsala University
Sweden
Phone: +46 (0)18 471 4315
Private: +46 (0)733 924 657
http://www.farmbio.uu.se/research/researchgroups/pharmacometrics/
Quoted reply history
From: owner-nmusers_at_globomaxnm.com <owner-nmusers_at_globomaxnm.com> On Behalf Of Rik Schoemaker
Sent: 29 August 2019 15:33
To: nmusers_at_globomaxnm.com
Subject: [NMusers] Lognormal survival in NONMEM?
Dear all,
Playing with repeated time to event models, I run into the issue that simple diagnostics for a single time to event outcome suggest that constant hazard and Weibull models are not very appropriate. The lognormal model seems to provide a very nice fit; compared to a constant hazard, the hazard is suggested to be higher in the beginning and then significantly lower at later times.
I have not seen any implementations online: does anyone know if the lognormal survival function can be implemented in NONMEM, and/or can anyone suggest alternative approaches? Some time-varying function to modify the hazard?
Any and all suggestions appreciated!
Kind regards,
Rik
Rik Schoemaker, PhD
Occams Coperatie U.A.
Malandolaan 10
1187 HE Amstelveen
The Netherlands
http://www.occams.com
+31 20 441 6410
rik.schoemaker_at_occams.com<mailto:rik.schoemaker_at_occams.com>
[cid:image001.png_at_01D55E80.EB744C70]
Nr du har kontakt med oss p Uppsala universitet med e-post s innebr det att vi behandlar dina personuppgifter. Fr att lsa mer om hur vi gr det kan du lsa hr: http://www.uu.se/om-uu/dataskydd-personuppgifter/
E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: http://www.uu.se/en/about-uu/data-protection-policy
(image/png attachment: image001.png)
Hej, see example code below.
Ina Frobel also applied an empirical hazard model (Frobel et al) to catch the
pattern, see the model in DDMoRe model repository, with link to publication.
http://repository.ddmore.foundation/model/DDMODEL00000065
$PROBLEM Time to first event data
$SUBR ADVAN=13 TOL=9
$MODEL COMP=(HAZARD)
$PK
SIGM= THETA(1)*EXP(ETA(1))
MU=THETA(2)
$DES
DEL= 1E-12
TIM=T+DEL
LNT = LOG(TIM)
X1 =(LNT-MU)/SIGM
PDF= EXP(-1/2*(X1**2))/SQRT(2*3.14159265)
DADT(1)=1/(TIM*SIGM)*PDF/(1-PHI(X1))
$ERROR
CHZ = A(1)
SURX = EXP(-CHZ)
DELX = 1E-12
TIMX=TIME+DELX
LNTX = LOG(TIMX)
X1X =(LNTX-MU)/SIGM
PDFX= EXP(-1/2*(X1X**2))/SQRT(2*3.14159265)
HAZNOW=1/(TIMX*SIGM)*PDFX/(1-PHI(X1X))
Y=SURX
IF(DV.EQ.1) Y=SURX*HAZNOW
$THETA (0,1) ;SIGMA ;1 SD of the log normal distribution
$THETA (0) ;MU ;2 Mean of the log normal distribution
$OMEGA 0 FIX ;OM1 ;1 ;Only to tell NONMEM that each ID has
multiple rows
$ESTIM MAXEVAL=9999 METHOD=0 LIKE SIGL=9 NSIG=3 PRINT=1 MSFO=msfb1
BR, Siv
Siv http://katalog.uu.se/profile/?id=N96-5738, PhD
Researcher
Dept of Pharmaceutical Biosciences
Faculty of Pharmacy
Uppsala University
Sweden
Phone: +46 (0)18 471 4315
Private: +46 (0)733 924 657
http://www.farmbio.uu.se/research/researchgroups/pharmacometrics/
Quoted reply history
From: [email protected] <[email protected]> On Behalf Of
Rik Schoemaker
Sent: 29 August 2019 15:33
To: [email protected]
Subject: [NMusers] Lognormal survival in NONMEM?
Dear all,
Playing with repeated time to event models, I run into the issue that simple
diagnostics for a single time to event outcome suggest that constant hazard and
Weibull models are not very appropriate. The lognormal model seems to provide a
very nice fit; compared to a constant hazard, the hazard is suggested to be
higher in the beginning and then significantly lower at later times.
I have not seen any implementations online: does anyone know if the lognormal
survival function can be implemented in NONMEM, and/or can anyone suggest
alternative approaches? Some time-varying function to modify the hazard?
Any and all suggestions appreciated!
Kind regards,
Rik
Rik Schoemaker, PhD
Occams Coöperatie U.A.
Malandolaan 10
1187 HE Amstelveen
The Netherlands
http://www.occams.com
+31 20 441 6410
[email protected]<mailto:[email protected]>
[cid:[email protected]]
När du har kontakt med oss på Uppsala universitet med e-post så innebär det att
vi behandlar dina personuppgifter. För att läsa mer om hur vi gör det kan du
läsa här: http://www.uu.se/om-uu/dataskydd-personuppgifter/
E-mailing Uppsala University means that we will process your personal data. For
more information on how this is performed, please read here:
http://www.uu.se/en/about-uu/data-protection-policy
Dear Rik,
There are very nice code examples (for NONMEM) in this poster-material from
Nyberg et al.:
https://www.page-meeting.org/pdf_assets/404-Poster_PAGE%20_2014_tte_sim_joakim_nyberg_with_code.pdf
https://www.page-meeting.org/pdf_assets/404-Poster_PAGE%20_2014_tte_sim_joakim_nyberg_with_code.pdf
These include the log-normal distribution, as well as Gompertz and Weibull.
Best regards
Jakob
Jakob Ribbing, Ph.D.
Senior Consultant, Pharmetheus AB
Cell/Mobile: +46 (0)70 514 33 77
[email protected]
www.pharmetheus.com http://www.pharmetheus.com/
Phone, Office: +46 (0)18 513 328
Uppsala Science Park, Dag Hammarskjölds väg 36B
SE-752 37 Uppsala, Sweden
This communication is confidential and is only intended for the use of the
individual or entity to which it is directed. It may contain information that
is privileged and exempt from disclosure under applicable law. If you are not
the intended recipient please notify us immediately. Please do not copy it or
disclose its contents to any other person.
Quoted reply history
> On 29 Aug 2019, at 15:33, Rik Schoemaker <[email protected]> wrote:
>
> Dear all,
>
> Playing with repeated time to event models, I run into the issue that simple
> diagnostics for a single time to event outcome suggest that constant hazard
> and Weibull models are not very appropriate. The lognormal model seems to
> provide a very nice fit; compared to a constant hazard, the hazard is
> suggested to be higher in the beginning and then significantly lower at later
> times.
>
> I have not seen any implementations online: does anyone know if the lognormal
> survival function can be implemented in NONMEM, and/or can anyone suggest
> alternative approaches? Some time-varying function to modify the hazard?
>
> Any and all suggestions appreciated!
>
> Kind regards,
>
> Rik
>
>
>
> Rik Schoemaker, PhD
> Occams Coöperatie U.A.
> Malandolaan 10
> 1187 HE Amstelveen
> The Netherlands
> www.occams.com http://www.occams.com/
> +31 20 441 6410
> [email protected] <mailto:[email protected]>
>
> <image001.png>
Dear Siv, Jonathan, and Jakob,
Thank you all for your wonderful and insightful replies; great to see the
NONMEM forum alive and kicking 😊. Now I will try and implement to see if this
solves my issues!
Kind regards,
Rik
Quoted reply history
From: [email protected] <[email protected]> On Behalf Of
Jakob Ribbing
Sent: 29 August 2019 16:18
To: [email protected]
Subject: Re: [NMusers] Lognormal survival in NONMEM?
Dear Rik,
There are very nice code examples (for NONMEM) in this poster-material from
Nyberg et al.:
https://www.page-meeting.org/pdf_assets/404-Poster_PAGE%20_2014_tte_sim_joakim_nyberg_with_code.pdf
These include the log-normal distribution, as well as Gompertz and Weibull.
Best regards
Jakob
Jakob Ribbing, Ph.D.
Senior Consultant, Pharmetheus AB
Cell/Mobile: +46 (0)70 514 33 77
[email protected]<mailto:[email protected]>
http://www.pharmetheus.com/
Phone, Office: +46 (0)18 513 328
Uppsala Science Park, Dag Hammarskjölds väg 36B
SE-752 37 Uppsala, Sweden
This communication is confidential and is only intended for the use of the
individual or entity to which it is directed. It may contain information that
is privileged and exempt from disclosure under applicable law. If you are not
the intended recipient please notify us immediately. Please do not copy it or
disclose its contents to any other person.
On 29 Aug 2019, at 15:33, Rik Schoemaker
<[email protected]<mailto:[email protected]>> wrote:
Dear all,
Playing with repeated time to event models, I run into the issue that simple
diagnostics for a single time to event outcome suggest that constant hazard and
Weibull models are not very appropriate. The lognormal model seems to provide a
very nice fit; compared to a constant hazard, the hazard is suggested to be
higher in the beginning and then significantly lower at later times.
I have not seen any implementations online: does anyone know if the lognormal
survival function can be implemented in NONMEM, and/or can anyone suggest
alternative approaches? Some time-varying function to modify the hazard?
Any and all suggestions appreciated!
Kind regards,
Rik
Rik Schoemaker, PhD
Occams Coöperatie U.A.
Malandolaan 10
1187 HE Amstelveen
The Netherlands
http://www.occams.com/
+31 20 441 6410
[email protected]<mailto:[email protected]>
<image001.png>