In practice, one might use the model obtained at the end of section II to assign probabilities of death while in ICU to new patients as they are admitted to ICU and use the estimated probabilities along with other considerations to determine when to discharge these patients.

1. Use the current model to compute an estimate of:
2. A high-risk patient has an estimated probability of death greater than . List the top 10 high-risk patients sorted in decreasing order according to their estimated probability of death. From left to right, with 1 line per patient, your list should include the values for: or the other variables included in your final model. Use the column headings ID, AGE, SEX, etc. rather than . Are you surprised by anything on this list? What is the value of the response variable for the people in the list ?
3. For your final model, plot as a function of age, keeping other variables fixed at some level. Compare different curves and comment on the ones you find to be informative. For example, you may want to plot as a function of age when the levels of the other variables are low. This would correspond to relatively healthy patients. You may also want to try getting for different values of LOC (for example), and try plotting them versus age on one plot. Please indicate the levels you used for the different factors.
4. Find the 95% confidence interval for the relative risk between two ages keeping all other variables fixed at some level. Specifically, use ages of 60 and 80. What are the estimates of and ? Interpert these 3 quantities. Explain how you calculate the standard error, and what distibution you used. Indicate the levels of the different covariates.