Ready(?) for Trondheim

After the exam madness, I'm about to leave sunny England and off to relatively frosty Trondheim to attend and give a talk at the Second Workshop on Bayesian Inference for Latent Gaussian Models with Applications. Looks like a very interesting small conference, with lots of good talks.


Of course, I haven't finished preparing mine (but I'm unusually nearly there). I'll talk about my work on hierarchical models for data on In Vitro Fertilisation (IVF) and pre-implantation genetic screening (PGS). The main issue is that while we (or rather "they") are very good at fertilising eggs in a dish, we (again: they) are not so good at actually making the resulting embryo implant in the mother's womb, which means that the actual pregnancy rate is overall still about 30$-$40%.


The main problem is that once the eggs have been collected and fertilised, they are followed up for a few days after which the embryologist looks at them and decides which ones "look good"; these are then put back in the womb and hopefully develop in a pregnancy. But this of course does not account for potential problems in the genetic structure of the embryos, which are invisible to the naked eye. PGS is a complex genetic test that is able to detect potential chromosomal abnormalities in the embryos, thus making the choice of which are the "best" more informed. 


The theory that we are testing is that shorter telomere (some kind of protective cap sitting at the end of the chromosomes) are associated with chromosomal abnormalities. The nature of the data is obviously hierarchical, because normally we observe different cells, each of which belongs to one embryo, each of which is donated for research by a couple of parents. However, in this field there aren't many papers using appropriate methods to account for the implied correlation among the observations (in fact, I think there aren't any at all).


Our work seems to suggest that there is indeed some association between the length of telomere and chromosomal abnormalities $-$ especially for embryos that are not well developed. 

For "minor chaotic" embyros, which are as good as it gets in this setting, there doesn't seem to be any (or at least hardly any) effect of the length of telomere. In a way, even for the more problematic embryos (up to "uniformally abnormal", which are really screwed up) the effect is not very large and the probability of testing positive at PGS is not very much affected by the length of telomere. But for the intermediate embryos, the chance of testing positive to PGS is actually very much affected by telomere's length.


There are then potentially very important implications for the underlying science and for the management of IVF patients.


I'll post some more when I have finished the presentation. Also, we are working on a paper (just need to finalise a couple of details and re-submit it after having pleased the referees' thirst for modifications...)

England is the new Spain

Pizza, gelato and motorbike with no jacket. 


Can the weather be any better? (as Chandler Bing would put it)

Eurovision again

Looks like I've not got many new topics these days, but here's an update on the Eurovision contest paper. 


Yesterday, we saw a shocking documentary on Panorama about the conditions in Azerbaijan (here, something related); specifically it showed how the "royal" family (which is not royal at all - technically they are a republic with free elections) are basically ruling the country as a proper family affair. Moreover, according to the report, they are using the contest (which they will host later this week as holders of the prestigious crown) as a wild propaganda tool.


Anyway, one of the most shocking bits was an interview to some guy (can't remember exactly who he was) who was arrested because he dared voting for Armenia, who are the historical enemy. He argued that he voted for them as a protest against the fact that state television blocked the broadcast when the Armenian act sung their song.


I went back to our preliminary analysis and produced this.

Azerbaijan is the country in white in the bottom right corner of the map. Darker colours indicate a higher propensity to vote for any of the other countries (ie from Azerbaijan to others), while lighter colours are an indication of lower propensity to do so.

You may recall that we are trying all sorts of models, accounting for "cultural" and "geographical" proximity - this one tries to include both. I think it's interesting that effectively all the countries in the former Soviet bloc (more or less all the rightmost part of the map) are coloured in solid dark grey, but Armenia (the light grey country bordering Azerbaijan).

The "effects" are pooled over the different clusters (regions/spatial proximity), which I think explains the fact that there is still some propensity for the voting pattern Azerbaijan $\Rightarrow$ Armenia. But the difference with all the other former Soviet bloc countries is quite large!

Bayes Pharma 2012 (again)

Julien has posted his impressions and comments on the conference on Christian Robert's blog. 


Agreed entirely, although, while I think INLA really is cool, I also believe that to have more than one software/algorithm available is absolutely a bonus - provided one know what one is doing (but again, see my comment somewhere else about the use of "one"). 


So, I don't think that we'll have problems of "duplications" with JAGS/BUGS or self-coded MCMC stuff!


PS: I love Julien's title - "Bayes on drugs"

Marking exam papers

I'm hopelessly in the dark land where you spend the whole of your time marking exam papers. And I burnt my tongue at lunch.


Doesn't get much worse than that...

We're on technorati

In a bold and reckless move to take over the cyberspace, I've registered the blog on technorati. This is now the 90502-nd most popular blog in the world $-$ quite impressive, right? And more importantly, the 1863-rd most popular in the subtopic "science" and even more impressively, the 8th (out of 9, to tell it like it is...) under the search "Bayesian statistics". 


The task is become no. 1 at least in the science category by the end of June (yeah: right!) 

HPV paper

Our paper on HPV vaccination has been accepted for publication in Medical Care (I believe it will appear shortly in the electronic version). I think it was a nice piece of modelling, which we did in a very large research group (we needed so many people, clinicians, economists, etc - although if one wanted to be quite cynical, one might say that only a few of us did most of the work... But then again, as Sheldon Cooper would put it: "Incidentally, one can get beaten up in school simply by referring to oneself as one")


But I digress! Here is a presentation that I gave last year, presenting the main findings of the paper, and here is a graph of what these are:

That's the base-case scenario which assumes making vaccination available for all 12 years old and comparing vaccination to current practice (screening every 3 years). The cost-effectiveness of vaccination is evident - it will produce a (relatively small) increase in the overall costs of management of the disease, but it will also improve the quality of life (and by the way, HPV is kind of nasty and causes all sorts of problems from genital warts to cervical cancer!).

Even if you relax some of the assumptions (eg discount rates), the bottom line does not change terribly and vaccination seems to remain good value for money. The model is based on published evidence, individual data and clinical expert opinions, which we, of course, have integrated using a full Bayesian model.

There are quite a few things we could do better (ie some of the parameters are at the moment considered as deterministic, which of course they aren't). More importantly, there is a crucial and complex issue with herd immunity (so if you vaccinate girls, then they are less likely to be infected, thus when they have sex with a boy he is less likely to be infected, and then when he has sex with another girl she is less likely to be infected, ad libitum).

We're working on this - but it's a bitch to model! 

Eurovision contest

Marta and I have been working for sometimes on a cool (as well as kind of very low impact on the scientific community and on the progress of human understanding of the world) model to estimate the voting patterns in the Eurovision contest.


Here in the UK, there's always a lot of furore and the underlying understanding that England don't win because the other countries gang up against them. OK: may be this is not a big issue any more, since Sir Terry has indeed quit; but it normally pops up again in the tabloids after every contest when the English representative (invariably) ends up in the bottom part of the table. (Incidentally: may this have to do with the fact that a) the contest is quite crappy anyway; and b) England keep sending some weirdo up against a bunch of hot Eastern European girls?).


Anyway, long story short: we've run a few models. Basically the idea is to use a Bayesian hierarchical structure to investigate whether voting is driven by social and/or geographical patterns. Our main model includes structured cultural effects (identified by the "region" to which each voter belongs to), as well as geographical effects (modelled using a spatial structure, so that neighbouring countries tend to be more correlated).


We haven't quite had time to completely figure out what is actually going on. There definitely is an element to it, but rather than bias against countries, we seem to see some (natural?) inclination to vote culturally and geographically (not very exciting, you may say... well, at least it does make sense, right?).


I'll post some graphs when I think I know a bit better what's the punchline. The paper is on the way as well (but of course, I'm terribly late on the tight schedule that General Marta has prepared $-$ she probably won't be very happy about me spending time blogging about it, rather than doing it...)

Submitting a package to CRAN

The process is relatively smooth, I think - especially if you start off with a Linux machine. However, when I had to compile the package and documentation for BCEA, I still had a couple of blips. 


The first problem was in the documentation. R tries to automatically compile a LaTeX file which is derived from the Rd (documentation files - a bit of a pain in the arse to do, but helpful nonetheless). However, for some reason I was constantly getting a warning and the resulting pdf file was not correct (ie it would not have the necessary hyperlinks). 


After much searching on the web, I figured out that the problem was the absence of the LaTeX package texinfo. Funny thing is that when I tried to compile the .tex file using the command pdflatex file.tex it was working OK. So I think there's a problem in the way in which R manages the LaTeX compilation. On top of this, you'd need the LaTeX package inconsolata (otherwise the font for the code would look weird).

Book submitted

The book has been submitted to the editor. Now I only need to fill the marketing questionnaire - and I have a feeling it's going to be as (if not even more!) difficult...

Amazing EPL

I have to admit I've always had a soft spot for Roberto Mancini, as he was my childhood hero, while he was playing for Sampdoria. So, kind of happy for City (although they're not my favourite team, at all).


A bit of blucerchiato (check at 00.25sec) in this title, at least... 

BCEA on CRAN!

Finally, I got round to find some time to work out all the problems in compiling the BCEA (Bayesian Cost-Effectiveness Analysis) package.


I developed it as part of the work for the book. In a nutshell, what it does is the following: first, you need to specify, code and run a Bayesian model for your health economic problem. This may be based on individual level data, or be a decision-analytical model. Typically, this would be coded in JAGS/BUGS and linked to R using the usual libraries (R2jags/R2WinBUGS, etc). Either way, you end up with simulations from the posterior distributions of some suitable parameters, which you can combine in order to produce suitable variables of cost ($c$) and clinical benefit ($e$). 


At this, point, you load the library BCEA and a set of functions becomes available to produce standardised post-processing and economic analysis. I have coded some specific methods to produce plots and summary tables. Most of it is really simple stuff, but some are quite cool (IMHO!).


I'll produce more examples and try to post them online, here.

Bayes Pharma 2012

Earlier this week, I went to the third edition of Bayes Pharma, which turned out to be quite an interesting short conference. Most people in the audience probably were from the industry (although there was a very nice balance and quite a few people were from academia). The tone of the talks was, of course, pretty much of the Bayesian persuasion and the quality was very high, I thought!


Anyway, here is my (invited - thanks again for that, Julien Cornebise!) talk. I thought it went down pretty well and got quite a few interesting questions. The point of the talk is that while standard cost-effectiveness analyses assume perfect substitution (of the non cost-effective interventions, in favour of the most cost-effective one), in practice, quite often markets are not so elastic and some of the sub-optimal interventions are still available, thus leading to market inefficiencies. 


Tools like our beloved Expected Value of Information could be still used to quantify the loss in efficiency, like so:

The grey area indicates the loss in efficiency (ie the increase in the expected value of information, or to put it another way, the increased impact of parametric uncertainty on the optimal decision, given by the fact that non cost-effective alternatives are available on the marked).


Hope you'll like it!


PS Aachen was very nice too!