|
Abstract:
|
Often a clinical question is raised on how an observed pattern or variations in relevant prognostic factors may have effect individually on survival or on clinical outcomes during a period of time.The primary thrust of this paper is to address such a clinical question while identifying the variations in clinical outcomes with a recursive process using an Adaptive Random Jump Model. Adaptive filtering with intrinsic conditional autoregressive models(ICAR)are often used. Conventionally various rigorous specifications of ICAR models within a Gaussian Markov Randolph Fields(GMRF) scheme are used to detect changes in the mean of a short-run process. Alternatively, making informed decision with sequential information process over time can be considered using an Adaptive Markov Random Jump Model. It updates the recursive transition probabilities under this model being aligned with a Bayesian approach. Thus, it captures the variability and effect of endogenous factors over time. In a case study, using this alternative proposed model on hematologic malignancy in patients with multiple myeloma,the impact of clinically significant predictors on median survival has been explored.
|
