Hi experts! Have anybody experience in including an R-program as part of a larger system? In Matlab there is a toolbox that converts a m-script into C-code. One application in mind is that I do the model building in R, for estimating the risk for cancer based on clinical measurements. When the model is ready, a small R-program can simulate the model to estimate the risk for a new patient. The idea is that a doctor gets the measurements for the patient sitting in his office. Then he goes to Internet and types in the test measuremnets and gets the estimated risk. Look at www.finne.info for an early version to get the idea. There I developed the model in Matlab and converted it to Excel. Don't use the results! Much better are available in R! There are many more applications that need a higher degree of integration. Than you in advance. Ralf Finne SYH, University of Applied Sciences Vasa Finland
Ralf Finne wrote:> Hi experts! > > Have anybody experience in including an R-program > as part of a larger system? In Matlab there is a toolbox > that converts a m-script into C-code. > One application in mind is that I do the model building in R, > for estimating the risk for cancer based on clinical measurements. > > When the model is ready, a small R-program can simulate > the model to estimate the risk for a new patient. The idea is > that a doctor gets the measurements for the patient sitting in his > office. Then he goes to Internet and types in the test measuremnets > and gets the estimated risk. > Look at www.finne.info for an early version to get the idea. > There I developed the model in Matlab and converted it to Excel. > Don't use the results! Much better are available in R! > There are many more applications that need a higher degree > of integration. > > Than you in advance. > > Ralf Finne > SYH, University of Applied Sciences > Vasa FinlandThe Design package for R has a version for S-Plus. In S-Plus you can use its Dialog function to automatically create a GUI for getting predicted values from a series of fitted models. We will be working on an R version but it will publish the model to a web server. Frank -- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University
On Fri, 2007-04-27 at 20:29 +0300, Ralf Finne wrote:> Hi experts! > > Have anybody experience in including an R-program > as part of a larger system? In Matlab there is a toolbox > that converts a m-script into C-code. > One application in mind is that I do the model building in R, > for estimating the risk for cancer based on clinical measurements. > > When the model is ready, a small R-program can simulate > the model to estimate the risk for a new patient. The idea is > that a doctor gets the measurements for the patient sitting in his > office. Then he goes to Internet and types in the test measuremnets > and gets the estimated risk.We are in a similar situation and the approach we've used is to keep R as a backend computation engine (using Rserve) and have a web service interface to it. Thus when we develop a predictive model, we dump it on the backend and have a web service interface to it. With this interface we can then write a web page client or even include the model into custom code (command line or GUI). You can see some examples at: http://www.chembiogrid.org/cheminfo/rws/ames http://www.chembiogrid.org/cheminfo/ncidtp/dtp http://www.chembiogrid.org/cheminfo/pkcell/ The first two use previously built predictive models. The last one is a piece of R code. Alternatively you could use RApache ------------------------------------------------------------------- Rajarshi Guha <rguha at indiana.edu> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- A sine curve goes off to infinity, or at least the end of the blackboard. -- Prof. Steiner