Our first test session with Turk Talk, where we use a human-computer hybrid setup for simple natural language processing to assist the assessment of decision making, went really well – much better than expected with very few technical glitches.
We plan to hold the next one in a couple of weeks where we will stress test both the Turker and the software, to see how much cognitive load can be handled and what needs to be improved to increase this.
At present, we are focusing on mental health conditions for our topics and material but the approach does not have to be clinical at all.
Watch this space!
OpenLabyrinth will be featured at the AMEE eLearning Symposium in a couple of weeks.
We have a short two minute video on YouTube for the eLearning Symposium show & tell: https://www.youtube.com/watch?v=b4b62thWFmM
AMEE is the Association of Medical Education in Europe. This year, the annual conference is being held in Glasgow: http://www.amee.org/conferences/amee-2015
We will also be giving a 15 minute paper on the new CURIOS video mashup tool, which is closely integrated with OpenLabyrinth.
See you all there!
Well that was interesting. I found a nice little web service called BatchGeo. You can load a bunch of IP addresses into the site and it plots them out on a world map. So we did this for the first 250 logins for our demo server at http://demo.openlabyrinth.ca - this is what we got:
Location of first 250 logins
It would be even more interesting to do a cumulative map of all 17,000 logins but I’m too cheap to pay the fee to BatchGeo. But this does give you an idea of how widely OpenLabyrinth is used.
The first virtual patient case that I was introduced to by my colleague, Rachel Ellaway, was the Sarah-Jane case written by Dr Jonathan Round at St George’s University London.
I was smitten. And I am not afraid (although somewhat ashamed) to say that the first five times I played this case, I killed Sarah-Jane. It is a beautifully crafted case – deceptively simple in both presentation and demeanour, yet really quite difficult to get right. I kept coming back, determined to save her.
Here was the power of serious games, pulling me in to solve the challenge… and yet it is all just text and a good story. No multimedia. No fancy doohickeys. Just a great narrative.
I have cited this case over and over when preaching about the power of the narrative. Now, we have successfully ported it across to OpenLabyrinth v3. Check it out on our list of exemplar cases at
or you can go straight to the case at
We have a breakthrough!
Haven’t you always wanted to use natural language processing in your virtual patient cases? Now we have two ways of doing this. OK, full disclosure here: we are not talking Watson level full AI stuff! But there are many times with a virtual patient case design when you don’t want to prompt the user about possible answers to a question and cue them into the correct answer.
Now there have been virtual patients… or rather a virtual patient that did NLP to an amazing degree. The Maryland Project in 2007 had very impressive language processing – you could type almost any question you wanted into it and it would provide a sensible answer. But the cost and the programming effort were huge and not at all scalable.
We have had some basic text processing capabilities in OpenLabyrinth for a while now. Very useful in limited situations. But it is a pain considering all the variations that a user might type and allowing for these in the logic rules.
Now we have Turk Talk. Based on the concept of the Mechanical Turk, where a human pretends to be a computer, we have developed an interface where a human facilitator can handle text input from up to 8 learners in a small group session. Interface is done and stable – going into research testing now.
If you are interested in a collaborative project working at something like this, contact us.
After two years of research and code testing, we are delighted to let you know about our progress with semantic indexing in OpenLabyrinth.
Much of this progress is due to the work of Lazaros Ioannidis at Aristotle University, Thessaloniki in Greece. It is being featured in workshops and presentations at MEI 2015 in Thessaloniki today.
So what the heck is semantic indexing? The foundation of Web 3.0, it allows discovery of content in new and interesting ways. For our authors and learners using OpenLabyrinth, it opens up powerful new search capabilities and data visualizations.
Chord diagram generated from case concepts
Imagine that you want a virtual patient case that addresses that common complaint seen in the emergency department: chest wall pain. How would you find this? With past methods, unless that phrase appeared in the title or descriptors for the case, you would be out of luck. Worse still, there are many synonyms and codes applied to this presentation: costochondritis, ICD9 = 786, Tietze’s syndrome etc.
Semantic indexing opens up the possibility of searching by the concept of chest wall pain, looking up such synonyms and coding in related vocabularies and ontologies that already exist.
More on this as we refine the tool. If you are interested in this research and its applications, contact us.
We had a couple of slightly glitchy releases of OpenLabyrinth in the last wee while. I am pleased to tell you that v3.3.1, just released on Github at https://github.com/olab/Open-Labyrinth is good and stable.
If you want maximum stability but don’t care about the latest features, v3.11, released a while ago is probably the most stable.
Thanks to all of you who continue to give us feedback and help us with testing new features and bug fixes.
Over the past month, we have been migrating our two University of Calgary OpenLabyrinth servers from elderly Dell rack servers to brand new virtual servers on the UofC server farm.
http://demo.openlabyrinth.ca is our test and demonstration server and is used by many around the world as a test bed, prior to setting up their own server. It always has the latest version of the software on it.
http://vp.openlabyrinth.ca is our own UofC production server. It runs a more stable maintenance environment.
I’m pleased to say that the transition to the virtual server farm went well with minimal glitches. However, today, there seems to be a wee glitch on the vm server farm at UofC so the demo server (and a couple of other unrelated servers) are down for the moment. They should be up again in a few hours.
We have a number of OLab servers hosted by the University of Calgary. UofC are making some major network changes over the holidays so these servers may be down briefly during these two days.
We hope that this does not throw off any major projects for anyone. The servers involved are