Monday, June 21, 2010

Medical Images on TRIP

Our latest medical image update has occured and we've now got 113,876 images in our index.

For those of you who've not used our medical image search before we return the four top medical images for a given search term on the main results page. We then allow users to 'See more images' and as you'll see from the image below we automatically import Google images to enhance our own offering. Altogether our medical images are starting to look pretty solid.


Monday, June 14, 2010

New server

Wow, I'm impressed!

Last week I posted that we're moving servers and TRIP is now on there. For a few days it'll redirect to www.tripdatabase.co.uk but that's only while we muck around with some DNS issues.

Why am I impressed?

It is super, super quick - amazingly so.

I'm very, very happy.

Friday, June 11, 2010

Moving servers

Just a quick note that we're in the process of moving servers. We've been using the same servers for a number of years and one in particular is getting ancient and needs replacing. So, instead of buying new servers we've decided to move to the industry leader - Rackspace.

Our sites will be sitting on a brand new, high specification server. We'll also get better backup and support and most importantly we won't get charged extra if we're well used (we currently get penalised if we exceed a particular bandwidth figure).

The transfer is currently happening with TRIP Answers sitting on the new server and the main TRIP Database should follow as soon as the whole database (90 Gb) has transferred over. Once on the new servers we'll need to alter something called the DNS (the internet's address book) and then everything will be fine and much faster. In the interim you'll note that certain addresses re-direct to temporary addresses on the new server (for instance www.tripanswers.org points to http://tripanswers.tripdatabase.co.uk/). This should be no reason for concern.

Wednesday, June 09, 2010

TRIP in the future

It's been a while since I last posted, it's been a busy reflective time at TRIP. We've been preoccupied with our new project called TILT as well as looking at our business model. At the same time we're questioning the 'worth' of search (which overlaps with our TILT work).

Looking at the business model is difficult at the best of times and this is especially difficult when the people at TRIP are not natural business people. We're good at innovating and developing products, but turning these into commercial products has always proved difficult. At the heart of this is how to make 'search' profitable. In the mainstream world of Google there is the advertising model which works wonderfully for them. However, our efforts at TRIP to use such a model have never really delivered, even with our hundreds of thousands of searches per month.

We've been in a few discussions with bigger companies recently (and these are ongoing) looking at working more closely with them to benefit from the better business brains. It's very early stages but we're hoping that more formal arrangements will be forthcoming, giving TRIP a firm foundation to carry on and develop. One of these developments is to move on from the current state of search.

I've posted a few times on the limitations of search but will briefly recap here. Imagine if search engines had not been developed and you got 10,000 clinicians in a room and asked them what product would they like to see if they had a clinical question. I would be surprised if a single one would say they wanted to see a screen of 10-20 results which may contain the answer to their question - based on searching with 1-3 search terms related to their clinical question. Yet, this is the current situation. While TRIP is a great search engine, search is limited.

The next stage of TRIP is to try and move on from the above paradigm and TILT is our approach. Without giving too much away, TILT seeks to extract clinically useful content from a wide range of sources. So, you may have a research article which contains multiple nuggets of evidence/knowledge. Why leave them hidden in the document? So, much of TILT is about deconstructing evidence into constituent parts. However, there is another very important layer above that. The layer is how to decide what nuggets are clinicially useful/relevant.

We're mid-way through piloting our approach and we're very close to being able to deliver on the above. How our approach is initially received and adopted will decide it's success and having decent partners to help market and approach TILT will be invaluable. We're also looking to content providers to work with us and we're currently in discussions with a few groups we'd like to have on-board at launch.

After over ten years working in search and information support for clinicians I'm getting really excited as TILT has the potential to move significantly closer to where we need to be in helping clinicians practice better healthcare.