brainly
Search...
Unlock all answers

marquezbell5168
06/14/2023
Medicine
College
answered • expert verified
From Collett, Modeling Binary Data". The treatment regime to be adopted for patients who have been diagnosed as having cancer of the prostate is crucially dependent upon whether or not the cancer has spread to the surrounding lymph nodes. Indeed, a laparotomy may be performed to ascertain the extent of this nodal involvement. There are a number of variables that are indicative of nodal inwolvement which can be measured without the need for surgery, and the aim of a study reported by Brown (1980) was to determine whether a combination of five yariables could be used to forecast whether or not the cancer has spread to the Iymph nodes. The five variables were: age of patient at diagnosis (in years). level of serum acid phosphatase, the result of an x-ray examination (0=negative. 1-positive), the size of the tumour as determined by a rectal examination (0*small, 1=large), and a summary of the pathological grade of the tumour determined from a biopsy (0mless serious, 1 more serious). The values of each of these five variables were obtained for 53 patients presenting with prostatic cancer who had also undergone a laparotomy. The result of the laparotomy is a binary response variable where zero signifies the absence of, and unity the presence of nodal involvement. The data are listed below. The problem here is to identify whether all or just a subset of these variables are required in a model that could be used to predict nodal involvement (nodal = 1 is nodal involvement and nodal =0 is no nodal involvement). a) State the type of regression model suitable for these data and why and the n fit the regression model to the nodal involvement data stored in pcancer −1 dta