Trip Database Blog

TRIP released a load of optimisations at the start of last week (22/23rd July). This resulted in the speed going from slow to acceptable/good. Giving it a time frame it has gone from 10+ seconds per search to around 2-3 seconds.

Since the roll out of the speed increase our traffic has jumped again. Yesterday we were searched over 52,000 times. Looking at the last 3 days we were searched 129,938 times (includes a Sunday). Compare that to the same 3 days in mid June, we had 104,285 searches.

Entirely sensibly, people like speed in searching and have better things to do than wait!

David Rothman has highlighted a delightful slideshare presentation (click here). I think I’ve managed to embed it below!

I still think there is a great benefit to health being made more social via the internet, by that I mean sharing experience. How this experience interfaces with evidence is another story. One thing that is pretty clear is that evidence takes second/third place to personal experience and to the experience of people we trust.

Anyway, enjoy the presentation…

Hot on my post on MedPedia comes the long-awaited (!) release of Google’s wikipedia-competitor – Knol. You can read a review via this ReadWriteWeb article.

There seems a number of articles on health topics (aimed at consumers) and the author of the article is clearer shown (alongside their credentials) – addressing the main criticism of wikipedia; in that the authors are known. Users can suggest alterations – but the author needs to approve. A few exampl articles:

• Cancer genetics
• Cataracts
• Mitral Valve Disease

This will be interesting to watch to see if people prefer Knol or Wikipedia. A few years ago Google was perceived in a much more friendly light than it is now – so will users flock to suggest alterations and improve the articles? I think the possibility of payments (via google adsense) might just swing it…

An interesting new development:

“MedPedia is a new project, currently in development, that will offer an online collaborative medical encyclopedia for use by the general public. In order to keep the content accurate and up-to-date, content editors and creators have to have an MD or a PhD. Several highly-esteemed medical colleges will be contributing content to MedPedia, including Harvard Medical School, Stanford School of Medicine, UC Berkeley School of Public Health, and University of Michigan Medical School. Medpedia is also receiving support from the National Institutes of Health (NIH), the Centers for Disease Control (CDC), the Federal Drug Administration (FDA) and many other government research groups. The content from these organizations will then be edited by MedPedia’s community of medical professionals.”

Click here for further details

I’m very happy with progress on the slow speed of TRIP – we’re no longer slow! I appreciate slowness is relative, but we’ve gone from 10+ seconds for the search results to be shown to between 1-2 seconds. Before it took and age and now it seems quick.

The basic problem was the way the system handled the snippets (the small pieces of text below the document title). These were identified as the bottleneck and so they way the results are returned has been changed to take into account this slowness. Previously, the results and snippets were generated and displayed. Now the results are displayed and the snippets dropped in when they’re ready. There are other potential areas of optimisation but we can consider these at our leisure!

Apologies for the slowness of the system to date and enjoy the new speed.

TRIP has been busy recently with 3 major projects due to launch by the end of the summer. The Spanish version of TRIP should be out very soon, TRIPanswers will be out by the end of August and TRIPcpd should be out by the end of September.

In addition, we’ll be rolling out a small upgrade to TRIP soon, with the ability to export records and e-mail to a colleague.

The TRIP speed issue is also being worked on with a final report from our web-people due soon.

So, while we’ve been busy, most of them are under control and nearing completion so we can start thinking about our next batch of work – an upgrade to TRIP. We’ve got a number of ideas for the upgrade, but we want your help. What would you like to see?

Please contact us (click here) with any ideas you may have. No matter how big or seemingly trivial, we want to hear.

Over to you to help improve TRIP

TechCrunch has just posted an article on Ologeez!. Ologeez! is an alternative front-end to PubMed (but has plans to include other databases as well) and mixed in with ‘social’ features.

Two things sprang to mind when reading this coverage:

• Alf Eaton already has a pretty impressive ‘version’ that has been around for years – HubMed. It was nice to see someone else highlighting this in the ‘comments’ section.
• Aside from HubMed there are loads of other PubMed ‘versions’ – yet it was this one that gets a mention in TechCrunch – a Stanford bias perhaps?

You can see Ologeez! by clicking here. It’ll be interesting to see if it succeeds, I have my doubts. Our playing with this sort of thing (courtesy of gwagle) was an interesting failure. As mentioned above there are numerous other websites offering social functionality coupled to PubMed, yet none are successful. I can’t see anything in Ologeez! that makes me think they’re particularly special. Perhaps the TechCrunch link will help (it certainly can’t hurt).

I suppose one day someone will crack it….

One quick thought, instead of having one site trying to monopolise this social space, why not get together a significant number of ‘players’ in the area – be it medicine, be it biomedical – and create a single space that all users can contribute to. Imagine if TRIP, Cochrane, Medline, NICE, CRD, eMedicine, BMJ etc etc all worked to one standard and created a single ‘space’ – now that would be powerful and powerfully useful. One to ponder.

Much is written about bias in relation to EBM. The overwhelming focus is on methodological bias in creating primary or secondary reviews. However, little is written about a bias that is arguably as important – search bias.

What’s the point of spending huge amounts of money creating beautifully crafted pieces of research for them to be overlooked by a clinician looking for robust information to inform their decision making process?

The TRIP Database uses an algorithm which is a compromise and I would never claim it was perfect. The debacle with the NLH’s search 2.0 highlights further problems. Medline is a nightmare to use. In short there is no perfect clinical search engine.

So what does that mean for a practicing clinician? Do we want them to search multiple databases? I suppose, ideally, yes. But I’m far from convinced it is practical. There is an onus on clinical search engines to improve and meet the needs of the users. But the important message is that bias is not just about research methodology, it applies in equal measure to how clinicians access the research.

Wow, I can really see just how much time the NLH Q&A Service took. It stopped answering questions at the end of last week and I’ve already managed to empty my e-mail inbox, the first time in years!

There have been a number of e-mails with new content waiting to go on TRIP, the most notable being the rather nice York Health Economics Consortium.

Talking of NICE (tenuous I know) the Darzai review (click here) has indicated an expanded role for NICE. It looks like they’ll be responsible for a new service/organisation NHS Evidence:

“NICE will manage the synthesis and spread of knowledge through NHS Evidence – a new, single portal, through which anyone will be able to access clinical and non-clinical evidence and best practice, both what high quality care looks like and how to deliver it.”

I can only surmise that the NLH will end up moving (again) to fall within NICE’s clutches. What that means for Q&A I wouldn’t like to say. I’m seeing Peter Hill tomorrow and that might shed some light on things.

The extra time on my hands has been spent harassing our web-developers and we’re fixing a number of niggling problems relating to advanced search, truncation, apostrophe’s. In addition – for some reason – the big 5 core journals (NEJM, JAMA, Lancet, BMJ, Annals of Internal Medicine) have disappeared from our search. These should be back soon.

TRIPanswers is progressing very well and the beta should be available within a fortnight. It should be as good as finished then but I just need to add a load of content. I’m hoping to launch with over 5,000 Q&As, reaching 10,000 by the end of 2008. I haven’t thought much past then. Each Q&A will be quality marked and have various other bits and bobs to transform Q&A; at least a little bit!

I’m also starting to research various semantic tools to improve various aspects of the TRIP site. I’m hoping to get some sort of system set-up to allow me to play around with these. I see these tools as allowing things such as semi-automated updating of Q&As, allowing natural language searching of TRIP and TRIPanswers. I even thought we could allow users to add text that they are interested in and use that as a basis to find ‘related articles’ within TRIP.

I’ve been a long-standing fan of PubMed’s related articles and getting this close to trying similar tools on TRIP is very exciting. I’ve often carried out searches of PubMed and found articles of interest only to find even more by using ‘related articles’.

Jumping back to TRIPanswers, if anyone wants an invite to the beta, let me know via http://www.tripdatabase.com/ContactUs.html