Graduate Student Seminar
Wednesday, April 15, 2015 · 11 AM - 12 PM
Session Chair | Mingyu Xi |
Discussant | Dr. Mathew |
Speaker 1: Marilena Flouri
- Title
-
Tolerance Limits for Cost-Effectiveness Analysis
- Abstract
- Cost-effectiveness analysis is a highly important methodology in the hands of policymakers of health technology, since it merges information about health outcomes and costs, and helps them to allocate their resources in the most gainful way. In order to quantify cost-effectiveness, several measures have been proposed in the literature. The most common used criteria are the Incremental Cost-Effectiveness Ratio (ICER) and the Incremental Net Benefit (INB). In this talk, we discuss criteria that can bring out features not captured by summary measures such as ICER and INB. Features such as: how does the difference between the costs compare with the difference in effectiveness? For a majority of the population, is the increase in cost too large compared to the effectiveness gain? These are clearly natural questions for a policymaker. In order to address these questions, we consider random variables motivated by the definitions of summary measures such as ICER, INB, etc., and discuss the computation of tolerance limits for the distribution of these random variables, using data from a 2-arm Randomized Clinical Trial (RCT) with patients being randomized to two treatments. In particular, inference on the medians and percentiles of these random variables will be addressed.
Speaker 2: Iris Gauran
- Title
-
Modeling Clustered Survival Data with Cured Fraction
- Abstract
- In modeling lifetime data, standard parametric theory assumes that all observations will eventually experience the event of interest if they are monitored for a very long period. While every unit starts as susceptible to the event of interest, a fraction of observations may switch into a non-susceptible group. A mixture cured fraction model with covariates is modified to incorporate random clustering effect to characterize the switch mechanism. Simulation studies and telecommunications data show that cured fraction models with random clustering effect perform better than their parametric counterpart in terms of predictive ability.