[R] nonparametric mixed model ?

[Troels Ring]

Dear friends. I hope you will bear over with a lengthy question. I have a problem of 76 patients on hemodialysis measured repeatedly once a month for nine months for methylmalonic acid and vitamin B12. The vitamin had been administered previously to these patients but was curtailed in the first part and then started again. There was a very clear response in that as vitamin B12 went down methylmalonic acid went up. This is as expected for people with diagnosed vitamin B12 deficiency but these patients had rather high vitamin B12 concentrations and most people wouldn't believe them to be vitamin B12 deficient. We expected the relationship between B12 and MMA to be nonlinear in that when B12 was high enough, MMA wouldn't increase. We have modeled the response in Genstat using the new REML facility and fitted a spline which shows the expected asymptote and also a 1.st order time series. In R I have fitted a Gee model of the marginals also with a first order time series and found the expected very clear negative effect of B12 on MMA. What I would like also was to fit the nonparametric mixed model in some other environment that Genstat. Perhaps the sm covariance analysis might do it but would that accomplish a mixed model ? There is for a nonspecialist perhaps the problem with all the interesting ways in R to pick up the method you really need. Some kind of index to them all or search facility would be very helpful if possible, someday. 

Cheers
Troels Ring, MD
Department of Nephrology
Aalborg Hospital
Aalborg Denmark


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._