[R] fda modeling

[Troels Ring]

Dear friends - We have 25 rats, 14 of these subjected to partial removal 
of kidney tissue, 11 to sham operation, and then followed for 6 weeks. 
So far we have data on 26 urine metabolites measured by NMR 7 times 
during the observation. I have smoothed the measurements by b.splines in 
fda including a roughness penalty, and inspecting the mean curves for 
nephrectomized and sham animals indicate differences for several of the 
metabolites. Now the real idea is to use the NMR measurements to 
understand what goes on in the kidneys since we know the partial removal 
of kidney tissue will result in progressive damage in the kidneys - the 
nature of that is what we want to understand. We have a blood sample 
from the rats just prior to sacrifice, and the creatinine concentration 
there is a good proxy for "renal function". So the course of 
concentrations of the metabolites are thought to be valuable in 
understanding the physiology. Some of these are thought to be 
correlated. We have two groups where sham animals have better renal 
function than partially nephrectomized, but there is variation in both 
groups which is also interesting - some animals progress more rapidly 
after "the same operation"  than others - we would like to know why.
The data are available (eventually - the resulting blood tests still are 
missing) if anyone would like to have a look but the main issue is if it 
is at all feasible to make fda work on such a problem.
Best wishes
Troels Ring,
Nephrology
Aalborg,  Denmark