Bayesian statistics

Travel man

In the past couple of weeks, I have been doing lots of (mainly) work-related travelling — before you ask, that’s not me in the photo. But it’s not a (too) distant version of me either… Last week, I was in NYC to run a training course on Bayesian methods in health economics — kind of a shorter/condensed version of our summer school (incidentally, although the website is currently still pointing to last June’s edition, we’ve now finalised the dates for 2020.

Bayes in America

For the first time ever, the annual Bayes 20XX/Bayesian Biostatistics is going to be held outside Europe — in fact we’re going to be in Washington DC. It will be an unusual edition — the timing is also very different; normally we have the workshop late May/early June. But this year we’ve defered it to late September to more or less fit with other conferences happening at the FDA. We’ve made a big effort in creating stronger links with the regulators (and the Americans, in particular), so this is certainly a great result for our “Founding Father” Bruno Boulanger!

Short course: Bayesian methods in health economics

26-28 November 2019, ICON Plc, Stockholm office Instructor: Gianluca Baio (UCL) Lecture topics Day 1 - Introduction to Bayesian modelling and application to health economic evaluation Start End Topic Description 10:00 11:00 Lecture 1. Intro to Bayesian modelling The Bayesian paradigm - expressing uncertainty using probabilities. Overview of probability distributions for different types of quantity. Predicting data with uncertain parameters. Introduction to Monte Carlo sampling in BUGS.

To do list

-- Over the summer, I’ve had several interesting discussion with colleagues, working/using/trying out some of our packages, including BCEA, survHE and SWSamp. Joke Bilcke (to be honest, I’m not sure this is the best link to her webpage, but I hope it is!) contacted me to suggest a couple of changes in BCEA. In particular, she has pointed out that when doing an analysis for multiple comparisons (which in this context simply means that there are more than 2 interventions being assessed), then the standard analysis considers by default pairwise comparisons.

Health economics training in the Autumn

I’m more or less back full time from an interesting summer break — Marta and I decided that going to a conference two 7-hours flight away, with two children (7 and not even 2 years old) was, if not a good idea, something we ought to try. In the event, it did go well, I think — we spent a bit of time visiting Malaysia and then “attended” the ISI World Conference.

Huffing and puffing

I kind of feel like Mark in Peep Show, when he pretended he was fit and able to do jogging with Big Suze and after only a minute or so he ends up nearly vomiting… BUT: the rush in the past few weeks has been worth it, I think. First I did my talk at O’Bayes — I showed some of the work on VoI and while preparing my slides spending waaay too much time photoshopping Rev Bayes into a St Patrick’s crowd:

Tea party

There I am, sipping tea from my brand new highly personalised mugs! (Notice the cool logo as well as the link to the website). As a matter of fact, we’ll give these magnificent objects out as gifts the participants to our short course (that’s on Monday 8th July) - that of course should be enough to convince you to come (but there’re no more spaces, so that’s not the point…).

A couple of things...

Yes — there’s a couple of things I wanted to share. Firstly, our chapter on Bayesian Statistical Economic Evaluation Methods for Health Technology Assessment has just appeared in the Oxford Research Encyclopedia of Economics and Finance. I wrote this with “the Andreas” (Gabrio and Manca) — and I quite enjoyed the process. A lot of this is coming from stuff that I’d already worked on and written, but Andrea M has brought in quite a few nice little (but subtle) points, particularly from the wider economic perspective.


This is pretty exciting (I know… in a very nerdy kind of way…). So: Value of Information is one of the things I’m mostly interested in. Over and above the pleasant time on a personal level, the work we’ve developed with Anna and Ioanna has been very cool, I think — we’ve certainly had lots of fun. Anyway, during Anna’s PhD period, we’ve started to have several discussions and kind-of informal collaborations with a few of the people working on VoI in the UK — the usual suspects from the Statistical point of view (unsurprisingly, Mark, Nicky and Chris).

Too much?

Yes. _Definitely_ too much.