A guest blog from Paul Henderson of Microsoft partner Ascribe.

There is a lot of talk about Big Data. You can’t help but feel that a lot of the talk is about the opportunity to think differently about data to improve your position in whatever industry you might be in. Very little seems to be about actual execution. Who is radically transforming the way they work? Perhaps that isn’t the question, though. Perhaps it isn’t a bad thing that everybody’s talking, like the song says, and perhaps the question should be, what can we do differently if we think differently.


If you try to explain big data to the man on the Clapham Omnibus (for international readers, that just means the everyday man-in-the-street) you will most likely say that we are creating a lot of data, very quickly, from lots of different places and in lots of different formats as more of what we do gets recorded in digital format. You might give some eye-popping numbers about how big your big data is, but don’t expect him to rock back on his heels. However if you tell him that Obama worked out which members of the public might change their mind about who they would vote for, then what issues would make them change their mind and then directed content at them to tell them why he was the man for them because his views on said issue were aligned to theirs – that’s a story!  It’s not a mainstream story – not everybody is doing this kind of Jedi mind-trick, but it is possible if you think differently.

Change gear and think about health and social care now. That is my sector. We are awash with data, disproportionate to the amount of wisdom we are able to derive from it. We have device hungry hospitals, measuring everything, and increasingly we are deploying these devices to the home thereby exponentially growing the volume of data that can be harvested from devices. Healthcare data, by and large moves slow, whereas the business of healthcare moves fast – whether that is treating patients with immediate needs or responding to national pandemics. The reality is that most healthcare decisions, once you move away from the patient in front of you as a clinician, is that they are made in the rear-view based on metrics like length of stay, price and other Key Performance Indicators that have been around for a long time. True, we are making good progress on outcome measurement and KPIs that measure things that are valuable to the patient, but there is the rub. Patients value the things they tell you about. We understand the things we hear and write down – not in numbers but words.

Last year I saw a demo at Microsoft World Partner Conference that was thinking about sentiment analysis in a big, Big Data way. If that isn’t too many “big”s. They were thinking about tweets about actors and judging  who’s star was rising and who’s was waning.  I wasn’t that taken with the outputs of the study because everybody loves Judi Dench but I was taken with the process and the technology. So the worldwide health team and I embarked on a project to see what we could do, with doctors’ notes using the same technology stack that had been used in a marketing presentation. The results amazed me. Behind the talk and the slideware there is actually a clear technology stack delivering hard RoI (Return on Investment) for healthcare.

Working with our customer in Leeds, in the UK, we looked at data in their Ascribe Unscheduled Care system. They generate a million rows a month, just in their Accident and Emergency system, from .5m attendances per annum. As an order of magnitude. The hospital staff are brilliant. They offer exemplary levels of care at scale, all day everyday. However they do keep a lot of notes that are used at the point of care but not really mined for the gold that is hidden in them. We were able to map the notes to drug dictionaries and other clinical taxonomies so that we could look for, for example, outbreaks of infectious diseases and alcohol-related attendances. These are issues that bite both locally and nationally, as the enclosed case study shows. We built Windows8 apps, for slates, to show how we could collect textual data in a more economical way, pushed data from our Ascribe application to WindowsAzure running HDInsight, to process data at scale, our friends at Two10degrees worked with us to refine their Natural Language Processing technologies to parse the data, and we delivered the resulting analytics back to site using Excel 2013 advanced mash-up and visualisation tools. Sounds easy in a sentence. Well, actually, it wasn’t too tough to do in the 8 weeks we had available, because the technology worked pretty seamlessly.

The project was a proof of concept. The results aren’t ready for full clinical deployment yet, however the application we built is now being trialled all over the world (China, US, Singapore, Holland) and we are shortly to move to pilot in Leeds. The drivers for the international work are all the unique ability to process data at scale, data that comes from very humble beginnings of notes of visit transcribed in cubicles in hospitals but by the millions. So the talk is good. Think differently. Think about the art of the possible. Think about the use cases from projects from other industries. Think big but think small. The technology is now at a point where it is beyond your thinking and so you can be confident that you will be able to transform the way you work.

Paul Henderson
Head of Business and Clinical Intelligence
Ascribe