The first fortnightly BRAG meeting of the year will be held this Thursday 22nd February at 1 pm via Zoom/GP-Y801. This week we will have presentations by Yuxin and Luz.
Zoom Link: https://qut.zoom.us/j/7952714168
Yuxin’s Talk
Title: Bayesian spatial relative survival model to estimate the loss in life expectancy for cancer patients in Queensland
Abstract: To date, there have not been any population-based cancer studies quantifying geographical patterns of the loss in life expectancy (LLE) of people diagnosis with cancer. This absolute measurement of survival provides a fuller understanding of geographical disparities in survival outcomes for cancer patients. We propose using a spatial flexible parametric relative survival model in the Bayesian framework, which allows for the inclusion of spatial effects in hazard-level model components. This approach does not require information on the cause of death and allows complex and robust small-area estimation. The model was applied to breast, prostate, colorectal, and lung cancer data from the Queensland Cancer Registry across 528 geographical areas. The associated computer program scripts are available to support the understanding and implementation of our methodology to other spatial cancer modelling applications.
Luz’s Talk
Title: When to stop technology development for conservation of biodiversity?
Abstract: Technology development is becoming an essential investment to address the unprecedented crises our world faces. However, it remains unclear how best to allocate resources for technology development and deployment because of the uncertainties induced by Research and development (R&D): the outcomes of R&D and the effects of technology deployment are uncertain. We present a new approach to solve this problem using Partially Observable Markov Decision Processes. Our model is innovative because it simultaneously plans the development and deployment of a new technology and dynamically adapts to possible technology development or deployment failures. Our analysis reveals it is optimal to invest in technology development for a predetermined time-period before surrender. This is a key finding because it implies that decision-makers do not necessarily need to evaluate the feasibility of the R&D program every year to decide to pursue development. If the new technology is ready at any time during that optimal time-period, our model undertakes an adaptive deployment process to dynamically learn the best way to deploy that new technology. We find that this optimal time-period depends on the dynamics of the system under current management, the initial belief in technology feasibility and costs of technology development and deployment. We provide user-friendly guidelines building on an analytical approximation, which allows decision-makers to use our model without requiring expertise on how to solve a POMDP. We present our results with a case study inspired by the management of a collapsing ecosystem: the Great Barrier Reef, Australia
Bayesian Research & Applications Group 