While
I have envisioned a career related to statistics for several years, it
was during my undergraduate studies at university that I developed a
passion for the subject. My experience during the summer of 2009 when I
worked on a project, entitled 'A Divergence Weighted Independence Graph
analysis of data from the British Crime Survey', was my first taste in
applying the concepts I had learned at undergraduate study to a real
social economic problem rather than working on a hypothetical
question. This exposure to research not only helped me decide to commit
to further study but made me realise my interest in implementing my
theoretical knowledge to a real world problem.
At this moment in time I am learning
about subjects and areas new to me in
both Statistics and in particular Operational Research (OR). Having
enjoyed being a maths and stats undergraduate I am still new to the
world of OR, because of this an influential factor for me in selecting
the STOR-i program was the prospect of gaining a greater insight into
OR. Modules which I will by studying this year are:
-
Bayesian Inference
-
Computationally Intensive Methods
-
Computer Simulation
-
Extreme Value Theory
-
Forecasting
-
Likelihood Inference
-
Optimisation
-
Probability and Stochastic Processes
-
Training for Research and Industry
Current research
The NHS has been has been in charge of public health in the UK for over 50 years however, with steadily rising patient numbers and a shrinking budget the NHS are straining under the pressure of trying to maintain quality of service to patients while still meeting government set targets.
Improving the current situation provides a strong incentive for modelling hospital systems which can lead to better decision making and improve hospital efficiency. In particular enhancing the understanding of patient flows in these systems allows decision makers to identify potential operational problems such as bed blockages.
Healthcare modellers are faced by a number of different obstacles which include but not limited by:
the stochastic nature of arrival and service times,
uncertainty surrounding arrival rates and other elements of the model,
time dependent arrival rates; represented by non-homogeneous Poisson processes,
the network nature of hospital systems.
Incorporating these sources of variation is crucial for developing an accurate and useful model. But due to the variety and types of complexities found in healthcare systems constructing a suitable and robust model is far from straightforward.
Existing modelling techniques include both simulation based methods as well as more analytical techniques. These two approaches have their benefits and drawbacks but neither are necessarily all encompassing when it comes to modelling healthcare systems. Comparing the two methods, analysing how they integrate the different sources of variation can help to produce better models and improve the information available for decision makers.