Research in the Centre is currently focussed around nine distinct research themes relating to the application of Bayesian statistics in health economics.
Modelling Health State Preference Data
CHEBS has been looking at an alternative approach to modelling health state values using a (nonparametric) Bayesian method and comparing this to conventional regression based methods.
CHEBS' current focus to quantifying structural uncertainty in health economic decision models is to build on the discrepancy modelling research initiated by the Managing Uncertainty in Complex Models (MUCM) project.
Meta-modelling involves the use of statistical regression to learn the model input-output relationship. A popular regression method used in meta-modelling involves the use of Gaussian processes.
Bayesian Clinical Trial Simulation (BCTS)
Application of the Bayesian clinical trial simulation (BCTS) computational tool in planning clinical trials.
Statistical Issues in Clinical Trials with Economic Outcomes (such as missing data and censoring)
It is increasingly common for clinical trials to collect data on health outcomes and costs to evaluate cost-effectiveness. Cost data typically exhibit highly skew distributions. Two techniques whose validity does not depend on any specific form of underlying distribution are the bootstrap and methods based on asymptotic normality of sample means. Although these methods are technically valid, it is important to apply methods that recognise the skewness in cost data to maximise efficiency and avoid misleading inferences. CHEBS has developed a general Bayesian framework for cost-effectiveness analysis allowing for very flexible modelling of both cost and efficacy.
Modelling uncertainty in complex models
The MUCM project is developing a technology that is capable of addressing all sources of uncertainty in model predictions and quantifying their implications efficiently. It has the potential to revolutionise scientific debate by resolving the contradictions in competing models and making the uncertainties in model outputs transparent to modelers and end users alike.
Value of Information Analysis to Identify Research Priorities
Value of information analysis provides a mechanism for identifying which parameters in an economic evaluation are contributing most to the overall decision uncertainty so that research con be prioritised accordingly.
Bayesian Evidence Synthesis and Meta-Analysis
Economic evaluations require a synthesis of all available evidence into probability distributions for the input parameters. Bayesian evidence synthesis and meta-analysis provides a natural approach to representing uncertainty on input parameters. In addition, Bayesian methods allow a natural middle path to be taken between assuming that effects in different studies are known precisely or completely unknown, acknowledging that studies are related but not identical.
Elicitation of Expert Probabilities
It is not always possible to gather sufficient data to be able to ignore prior knowledge. Uncertain model input parameters are often assigned probability distributions entirely on the basis of expert judgments. The challenge is then to find appropriate questions to ask the expert in order to extract their knowledge, and then to determine a suitable probabilistic description of the variables we are interested in based on the information we have learned from them.