New funding for sustainability and take up

I'm pleased to announce that thanks to a great team effort over the last year and the endorsement and commitment of key biomedical research groups, both in University Hospitals Leicester Trust and nationally, we have been approved by JISC for sustainability and take up funding from July 2012 to end of March 2013. Further details to follow :)

i2b2 and CaTissue Integration

Last week Somaraja Surampudi and I successfully integrated i2b2 and caTissue on the eduserve cloud. We have prototyped an integration that models the way caTissue collects specimen data.

As i2b2 requires an ontology to create cohorts, we developed a nominal ontology that models the 4 main types of samples that caTissue collects, we have an ontology tree with four main nodes that are Cell, Fluid, Molecular and Tissue. Under each node there are further sub types such as DNA, RNA, FixedTissue.

From caTissue we have exported the following specimen related data :

  • Collection container
  • Collection protocol
  • Requested quality
  • Amount of aliquot collected
  • Date when sample was received
  • Date when sample was collected
  • Date when sample was frozen
  • Hospital location of sample
  • Which freezer and compartment the sample is located in
  • Barcode of sample
  • Protocol related to the sample

Currently we are in a testing phase and exports are run overnight.

So what can this integration achieve ? Well you could possibly build complex cohorts in i2b2 with defined dates, particular protocols, amount of specimen remaining, etc.

Malcolm Newbury's picture

Managing Research Data Hackday: 3/4 May

Great idea by Jonathan for the team to be present at the MRD Hackday in Manchester on 3/4th May. Soma and I left early on Thursday morning for Manchester, only to be greeted with a warm welcome from the Train drivers, who has succeeded in making us late by 2hrs for the hackday!

It was a well attended and facilitated group, consisting mainly of MRD programme projects, but also other projects, such as Dataflow and the leader of OSSWatch - Sander van der Waal.

From a Brisskit perspective the days were very productive. We took advantage of a 10 minute slot to present on behalf of the Brisskit project, our aims, progress and hopes for hackday, and I had some great discussions at lunch with Simon Hodson, Mahendra Mahey,John Nicholls over lunch.

The first day was spent in discussion groups, sharing ideas and problems around the need to create unique ID's, not just for patients, but also for their samples. We also discussed the need for de-identification techniques to enable data to be opened up for wider researcher access. We had some fast suggestions on how to deal with the ID generation problems and discussed at length the potential applications for an automatic pseudonomisation generator - not jsut for research purposes, but also for use in other online applications, where users would prefer their identities to be pseudonmised from the start of their engagement. Whilst we discussed this at great length, it was apparent that the solution would take longer than the time we had left at Hack day, so Soma and I decided to park this for Brisskit purposes and focus on working towards the integration of CiviCRm with i2b2 through BPEL.

At the end of the two days, we atmosphere was very conducive towards cutting code, and we did indeed manage to achieve a key breakthrough in being able to get past the stringent CiviCRm security and extract and play with Civicrm contact data within the openESB BPEL engine.

Next stop - i2b2 . . . .


Friday install fest

Today has been a pretty epic day for us; three separate bits of software all reached a major milestone - full installation into our vApp infrastructure. Whilst we have installed each separately on individual VMs before, this is the first time we have had them in our vApps. The vApps have their own private network so we can fence them off from each other for added security, this makes the installation process a bit more complex than if they were free floating VMs in the aether, but we solved all the issues today.

We now have i2b2 set up so it's visible with the workbench and the web client, caTissue working with the web client, and civiCRM working. We already have various pieces of the start and end bits of the integration for each bit of software, now we have all the software in the same vApp we can start plumbing them all together.

In the next couple of weeks we will have a demo version of everything running so you will be able to see the integration working and watch data flow from start to finish - watch this space!


Last week Dave, Saj, Soma and I went to Bristol for a CiviCRM training course. It was good fun and we learnt a lot :)

I don't think we have talked about CiviCRM much on the website yet, so let me give you a gentle introduction to what it is and how we plan on using it. CiviCRM ( is, as their website puts it:

"A web-based, open source, Constituent Relationship Management (CRM) software geared toward meeting the needs of non-profit and other civic-sector organizations."

In terms of BRISSkit it is a nice bit of software that we can use to manage all the participants a research group has, and manage their recruitment and progress through studies. As such CiviCRM is quickly moving to the core of our software stack.

As an example of a hypothetical use case, imagine a BRU with many thousands of registered participants, these would live in CiviCRM as contacts. Each contact can have any sort of data assigned to it, but we think mostly demographic data is the way to go. Now imagine a researcher wants to start a new study that looks to see what colour hair everyone called Bob has, and what their blood type is. He would design the study (write the questionnaire in REDCap and sort out the collection protocol in caTissue), add it to CiviCRM (modelled as a Case in CiviCRM), then he can start recruiting into it. He would then do a search of the contacts for people called Bob, then open a case for each one. This would automatically pre-seed REDCap and CaTissue with empty questionnaires linked to each Bob. Within CiviCRM the task of filling in the questionnaire and taking blood would be marked as 'scheduled', once it has been done it will automatically get changed to 'complete'. In this way the researcher can keep track of the progress of the study in real time.

CiviCRM has a reasonable looking API that we are using to plug it into the rest of the stack, so in the near future we will be able to start showing off this functionality.

Advances in Medical Sciences

Several of the BRISSkit team are in London today, presenting progress on the project to the 'Advances in Medical Science' conference, - there are a couple of hundred scientists and policy makers discussing the necessary building blocks for innovation in biomedical science. Getting the balance right is really difficult, I think. We've found tools which we think offer a route to a robust infrastructure for science, but by no means do we think they are the only tools which could do that job. The long-term objective has to be to shift the agenda so that delivering integrated, service-oriented cloud applications in a SaaS methodology becomes the standard approach for clinical research applications, to the point that instead of buying applications, researchers expect to be accessing a research platform. To reach that point will take a massive culture change inside clinical research, and for policy makers to help by shifting the legal and ethical framework to facilitate that approach, while the technologists deliver tools which allow data to be shared as appropriate, experiments to be replicated more easily and scientists be able to access and manipulate research data, garnered from the NHS, in ways which respect patients' confidentiality, with the scale necessary to derive clinically significant results, and verify them quickly.

The tools we have adopted for BRISSkit point a number of techniques for doing this, and the development of open source tools is a model which is pitifully under-used in clinical research despite being ideally suited to the needs of medical scientists. If events such as today's #ams12 can help to build a community towards that objective then we'll have done a very useful day's work.

Malcolm Newbury's picture

Clinical Integration . . . and why its a vitally important issue for Brisskit to address

Brisskit users will have a similar business process to our own Cardiovascular BRU at Glenfield, and our chosen applications for CRM, Subject engagement (onyx/redcap) , tissue management (ca Tissue) and Data Mining/Analystics (i2b2) are designed to serve this clinical research process. However, differnt clinical studies will require evidence to be gathered from a variety of expensive clinical systems, which are under the control of hospital staff. MRI machines are an obvious example. They produce very large MRI image studies in DICOM format and are often used to measure the outcomes of drugs, and clinical devices. Laboratory information systems is another obvious example - these use HL7 and other structured formats to share data..

Now interfacing with these systems using the DICOM and HL7 formats is not easy. It can cost upwards of £50k to develop and test these integrations and they will always need to be tweaked and retested for different manufacturers.

In the old days, this integration was seen as another source of revenue to equipment manufacturers, so nobody was too interested in making this job easier.
The cost of integration was always borne by the 'new system' or 'new interface' sponsor. For clinical research this becomes a massive barrier, because the value of data extraction cannot be easily reused when a study is complete, and is often discarded after every study.

Nowadays, manufacturers are less keen on integration revenue and more keen to make their systems easily assimulatable into a client environment. This is why standards bodies like Integrating the Healthcare Environment (IHE) have picked up a lot of world wide interest over the past 10 years and why they are inevitably going to pick up a lot of interest at Trust level in the UK over the next few years.

As far as MRI imaging is concerned you only have to look at the wiki site of the Royal College of Radiologists (RCR) to realise that IHE XDS is just around the corner. Hence the adoption by Brisskit of the IHE XDS standard for clinical systems integration will be a very smart move indeed.

JISC branding

Found a really useful page on the JISC web site describing how they want us to refer to them and how to use their logo.

Drupal and wiki

Some notes on integrating our Drupal and Mediawiki sites.

Using drupal as a first class wiki platform

MediaWiki extension to use Drupal for login

Do we need Mediawiki syntax when we have WYSIWYG editors?

User accounts and authentication

I've enabled the OpenID module, which means people can use their Google and Yahoo identities.

To configure your Google identity, select OpenIDIdentities on your user page and enter the URL of the Google OpenID service []. If you aren't already logged in to your Google account, the Google page will prompt you for your Google username and password. Once you have logged, you will be passed back to the Drupal site with your Google OpenID identity added to your account.

Once you have added your Google identity you can use this to login to the Drupal site.

Select 'login using OpenID' and enter the URL of the Google OpenID service [] in the box.

You will then be passed to the Google login page which will prompt you for your Google username and password