7. In Example 13.10 (head and neck cancer), retaining the existing time partition, fit a random walk intercept model with the prior differentiated by treatment group. Second, redefine the partition to have equal intervals (e.g. of length 1 month or 2 months) and use a conditionally autoregressive (CAR) prior to fit the RW1 model. This avoids the need to re-centre the random walk parameters at each iteration.