Can we fit this model in nlme or lme4? Monday, March 28, 2011 8:10 PM

Ben and others -

Thank you for your reply!!
Currently, I am using the following WinBUGS code to fit the model.
The estimation is good, but it took too much time to do some simulation.
So I was wondering if there is a way to do this fast since I only need the MLE estimate.



	     model
          {
           for (i in 1:nSUB) {
		           s[i] ~ dnorm(0, tau.precision);
               	}

          	for(j in 1 : nOBS) {
            		p[j] <- (E0 + Emax*Dose[j]/(ED50+Dose[j])+s[sub_seq[j]]) * (1-exp(k*X[j])) / (1-exp(k*d));
              	Y[j] ~ dnorm(p[j], sigma.precisionind[j]);
                sigma.precisionind[j] <- sigma.precision / pow((1-exp(k*X[j])) / (1-exp(k*d)), 2);}

           E0 ~ dnorm(0,0.0000001);
           Emax ~ dnorm(40,0.00000001);
           k~ dunif(-10,-0.001);
           ED50 ~ dunif(0.0001,10000);
           sigma.precision ~ dgamma(0.000001,0.000001) ;
           tau.precision ~ dgamma(0.000001,0.000001) ;
           Sigma<-sqrt(1/sigma.precision);
           Tau<-sqrt(1/tau.precision);
         }
     

Best,
Haoda

From: Ben Bolker <bbolker at gmail.com>
Subject: Re: [R-sig-ME] Can we fit this model in nlme or lme4? Monday, March 28,	2011 8:10 PM
To: r-sig-mixed-models at r-project.org
Date: Thursday, March 31, 2011, 7:22 AM
Haoda Fu <fuhds at ...> writes:

Hi,

I was wondering if the following model can be fitted

by nlme/lme4. 

The difficult part is that the error term
is combined with other parameters.
The model that we need to fit is as below and the R

codes 

to generate data is provided as well. Many thanks!!!

# Y_{ijk} = (E_0 + Emax*dose_i/(ED_50+dose_i) + 
? S_{ij} +

e_{ijk})*(1-exp(k*X_{ijk}))/(1-exp(k*d));

## Parameters to be estimated E0, Emax, ED50, sigma_S,

sigma_e, kn

? Thank you for the clear example.

? But no, I don't see any way to do this in nlme/lme4
(or in any other R mixed modeling package).
? WinBUGS, AD Model Builder ...

? Happy as always to be corrected by someone more
knowledgeable.

? Ben Bolker

_______________________________________________
R-sig-mixed-models at r-project.org
mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models