Talk and Poster at the 2018 Research Students' Conference


This year, the annual Research Students' Conference was held in Sheffield between 24th and 27th July. This was during the extended heatwave, so the weather was fantastic (sometimes a bit hot to be inside!), and there were lots of great talks and posters on display.

The full title of the conference is the Research Students' Conference in Probability and Statistics, and as you might expect, the Royal Statistical Society (and particularly the Young Statisticians Section) featured prominently. There was some twitter activity too, see @RSC_2019 (they updated their name after the conference...) and #RSC_2018 for more details.

This conference came when I was about half-way through my PhD (it started in January 2017), so it was an opportunity to showcase some initial results. I had both a talk and a poster, on separate topics.

My talk was entitled "Dynamic survival models and generalised linear models for analysing and extrapolating survival data in health technology assessment". The abstract is given at the end of this blog. Briefly, it discussed the theory behind analysing survival data within the framework of generalised linear models (and its extensions). This allows for the use of very flexible models. The feasibility of this approach was demonstrated using two case-studies.
The live-talk wasn't recorded, but I did a recording a couple of weeks later, which can be found here. Slides are here.

My poster was on the topic of "Parameter estimation in dynamic survival models: current status, future directions". It reports the results of a pearl-growing search: the most popular methods were Markov chain Monte Carlo, and Linear Bayes. More details are provided here.

Abstract for the RSC talk:
When analysing survival data, there is often interest in parametric modelling of the underlying hazard function. This can provide insight into how the process evolves over time, and allows for predictions of the future (extrapolations). This requires flexible models so that estimates are not reflections of the assumptions imposed by parametric models. Traditional models for survival data have been criticised for being insufficiently flexible; an alternative is to use generalised linear models. This framework and its extensions provide very flexible models. Examples include dynamic survival models, fractional polynomials, and spline-based models. Despite its usefulness, this approach to analysing survival data is rarely used. This talk describes the models and presents example case-studies. Particular attention is given to dynamic survival models; as they are based on time-series methods it is argued that they are more suitable for generating extrapolations. Possible areas for further research are also identified.

Comments

Post a Comment

Popular posts from this blog

Article: Public involvement in English research

This recent article contains a wealth of interesting information on public involvement in health research. Hence I thought that it would be worth discussing some of them here, and adding my thoughts on what this may mean for methodological research. The full title of the article is "How embedded is public involvement in mainstream health research in England a decade after policy implementation? A realist evaluation" and it can be found here . As a side-note, the authors use the term 'patient and public involvement' (PPI), whereas I generally refer to 'public involvement'. I do so as, first I take the term 'public' to include 'patients', and secondly because my work (being methodological in nature) will apply to the public in general, not to specific patient groups. However, I appreciate that depending on the context either term would be appropriate. Arguments for public involvement. The authors identify two main arguments for public
Read more

NIHR Infrastructure Doctoral Research Training Camp.

Image
In July 2018 I attended the annual National Institute for Health Research (NIHR) Infrastructure Doctoral Research Training Camp. I thought that I’d share some thoughts and reflections on this summer camp, if anyone else is thinking of attending. What is the camp? As the name suggests, it is an annual training camp for students doing doctoral studies (such as a PhD. The 'infrastructure' bit means that you have to be working as part of the NIHR Academy. This includes a number of things, such as the research schools  and biomedical research centres . I am working with the  Health Economics and Outcome Measurement theme of the  Yorkshire and Humber  Collaborations for Leadership in Applied Health Research and Care (I should try fitting that onto a name badge one day...) hence I was eligible to attend. Attendance is by invitation only: you have to write a letter saying why you think you should go to apply. Luckily I was chosen to attend. The camp has a different them
Read more

Extrapolation in HTA: Background reading

The problem of trying to predict future outcomes has long been a problem in HTA. The below provide some key reading on extrapolation, from a variety of different perspectives. All of the below should be open access, but if you have any problems please contact me . NICE Decision Support Unit Technical Support Document 14 (2013) This is a detailed report into the methods available for the analysis and extrapolation of time-to-event outcomes (also known as survival data). It was authored by the NICE Decision Support Unit (DSU), who provide support to NICE and their HTA work (the DSU have a series of Technical Support Documents , which are required reading for anyone working in HTA). This report had two main components: a review of extrapolation methods currently used, and guidance on producing extrapolations. This guidance includes both a model selection process algorithm, and clarification points. Since its publication, this report has been highly cited and used, both in the literat
Read more