To read the full case study download it here

Big Data in Emergency Care
Project Overview
Impact of Project
• Leeds Teaching Hospitals (“Leeds”)
is one of the largest Trusts in the
UK, serving a population of 2.5
million across the two Emergency
Departments at St James’s Hospital
and Leeds General Infirmary.
• Big Data from past to present:
Leeds Teaching Hospital started
using Ascribe Symphony seven
years ago and Ascribe was able to
apply NLP to this data store.
Nothing has been lost and the data
has been successfully used to
illustrate and track valuable
historical trends, thereby also
providing benchmarks for future
• The Emergency and Unscheduled
Care clinical IT system used by
Leeds, Ascribe’s Symphony, collates
a vast amount of patient and clinical
data, however the limited
availability of data analysts led to
limited availability of reports and
sometimes, delays. Clinicians were
not able to place information
requests, assess their impact and
react appropriately in a time
effective manner.
• Leeds Teaching Hospitals took park
in a pilot project with Ascribe’s
Business Intelligence (BI) team,
who had developed a Big Data
Natural Language Processing (NLP)
solution in conjunction with Intel,
Two10degrees and Microsoft, in
order to extract and analyse
structured and unstructured data
collated inside Ascribe Symphony.
• The Trust is now able to understand
healthcare trends captured within
their ED such as alcohol consumption, areas where the Trust was losing
income, and the impact of new
services, which gave the Trust
the evidence to present to their
Commissioning Board to secure further support for the ED.
• Key trends were identified to enable
the Trust to allocate resource and
request support for times when the
ED was particularly busy i.e.
University Fresher’s Week.
• ‘Loss of income areas’ were
identified; NLP applied to
unstructured data in doctors notes
(e.g. where ‘CT scan’ was only being
recorded anecdotally). This helped
the Trust to identify where
investigation income was being
unclaimed and effectively lost.
• Impact of new services was enabled
through NLP; a new mental health
service was co-located with the
Emergency Department to help
improve speed and efficiency of
referrals. By searching for the name
of the new service, using NLP, the
Trust identified that the new service
was being referred to twice as many
times as the old service.
Why select Ascribe Business
• Award winning Business
Intelligence solutions from Ascribe
- Microsoft Partner of the Year 2012
in Public Sector Health, we have
proven experience in
demonstrating excellence in
innovation and implementation of
healthcare information solutions
based on Microsoft technology
• ‘Making the unknown, known.’
Ascribe’s BI enables healthcare
organisations to understand quality
and financial key performance
indicators instantly.
• Improve productivity - Ascribe BI
identifies drivers that will help
improve productivity and service
• Improve patient care and
experience - enabling clinicians and
analysts to identify and understand
patterns and trends in patient care
enables organisations to identify
areas of improvement and provide
additional support to improve the
patient experience.
Leeds Teaching Hospitals NHS Trust is
one of the biggest NHS trusts in the
country, with over 15,000 staff and a
£1billion turnover.
The Trust has two main hospital sites St James’s Hospital and Leeds General
Infirmary. The hospitals both have
Emergency Departments, of which
80,000 adults are seen at St James’
each year and 80,000 adults plus
40,000 child patients are seen at
Leeds General Infirmary.
The trust’s vision is to be a ‘locally,
nationally and internationally
renowned centre of excellence for
patient care, education and research’.
Their purpose is ‘to deliver safe,
effective and personal healthcare for
every patient, every time’, which is
why it is essential to ensure every
service, including the Emergency
Department, is performing to it’s best
The Big Data Challenge
Big Data is defined as the collection of
data sets that are so large and
complex that they become difficult to
analyse using traditional database
Leeds Teaching Hospitals has used
Ascribe’s Emergency and Unscheduled
Care clinical IT solution (Symphony)
across both of its Emergency
Departments (ED) for the last seven
years. In that time, vast amounts of
data have been input into the system,
however the Trust did not have the
resources to analyse data and provide
real time information upon request.
Pressures from healthcare bodies such
as CCGs and the TDA (Trust Development
Agency) were increasing as they
required more up to date data
on the Trust’s performance; also
Police and Freedom of Information
requests for information on factors
such as alcohol consumption and
crime reduction were regularly
coming in and the Trust could not
provide timely responses.
Ascribe’s Business Intelligence team
approached Leeds Teaching Hospitals
to take part in a pilot project which
would help transform their huge silos
of data into meaningful reports that
would provide clinical insights and
better inform their care decisions.
Big Data NLP - the background
The project team consisted of Ascribe,
Microsoft, Intel and Two10degrees;
Ascribe’s BI team extracted patient
data from the Symphony system and
transferred it securely to Microsoft’s
cloud infrastructure.
From the cloud, the data was
processed by Two10degrees who
provided the NLP software; Ascribe
and Two10degrees worked together
to analyse the data, on Microsoft’s
Hadoop distribution platform,
HDInsight. The data was then relayed
back to Leeds Teaching Hospitals via
their on-premise data warehouse
which is held on HP servers powered
by Intel processors.
Essentially, NLP analyses free text
field data, extracts the information
and identifies key words within the
free text. After applying context and
then quantifying the data, it outputs
structured data that can then be cross
referenced against other sets of
structured data such as counts of
attendance, KPI breaches and
demographics using a link to turn the
data into patient-identifiable records
for use at point of care.
Case Studies
Ascribe worked with Leeds Teaching
Hospitals to identify some case studies
they could work on to illustrate the
potential of this Big Data NLP project.
Seven years of patient data was used
to create the following three case
1. Alcohol consumption and trends
in population
Leeds Teaching Hospitals wanted to
look at the impact of alcohol trends in
resourcing the ED. This data was not
kept within structured fields in Symphony,
it was kept in the free text (eNotes)
section. Alcohol wasn’t recorded
as a diagnosis as this is usually a
precursor to the actual diagnosis i.e.
head injury or fracture. Therefore
this was one of the hardest pieces of
information to record and extract.
Leeds Teaching Hospitals used NLP to
search for terms associated with
alcohol that were used in anecdotal
notes in the ED. This required a
sophisticated tool that looks at words
in context, for example ‘offered a drink
of water in ED’ would not be picked up
as the NLP recognised it was not in the
context of an alcoholic drink.
The NLP results were able to show
trends over weekends, amongst
certain age groups particularly
students, and Microsoft Bing Maps was
used to cross reference the
unstructured data with the structured
data i.e. postcode, time of admittance
etc. Leeds Teaching Hospitals wanted
to look for prevalent postcodes to
identify hotspots where alcohol related
attendances came from; by creating a
heat map with Bing Maps, the data
identified that LS6 was a high trend
area. LD6 is the student area of Leeds.
The ED was particularly interested in
the ‘Fresher’s Week affect’ – the impact
was a fivefold increase in August/
September – this was really important
for the Trust to be able to prove,
enabling additional resources to be
utilised during that time, providing a
redesigned service to meet patient
requirements and provide a better
These requests, coupled with the Hunt
report’s goal of paperless hospitals by
2018, meant that Leeds Teaching
Hospitals needed to find a solution to
transform their data into practical
information within an appropriate
timeframe. The solution was to apply
healthcare informatics.
Ascribe Symphony in use at the Emergency Department
2. Accurate capture of radiological
Leeds Teaching Hospitals was made
aware by a Capita report that the ED
was potentially losing a significant
proportion of income each year by not
fully recording all the treatments and
investigations they could claim income
from providing. Following research into
Ascribe Symphony, it was identified
that some investigations weren’t being
recorded in structured data fields, such
as CT scans for head injuries, but they
were being recorded in free text notes
By interrogating the unstructured data
field, the NLP tool identified that in
over 90% of records where the term
‘CT’ was found, it was only found
within the unstructured notes fields.
This highlighted a flaw in workflow
that has now been rectified.
The ED has now improved its billing and
income processes, ensuring work is
charged accurately and appropriately.
3. Implementation of the Acute
Liaison Psychiatry Service
Leeds Teaching Hospitals wanted to
improve the responsiveness between
ED and Mental Health services
referrals - therefore the Acute Liaison
Psychiatry Service was formed and
co-located with the ED to enable
closer patient working and enable
the ED to get people the right care
quickly. The ED wanted to measure
the impact of this new service.
Using NLP, the ED used search terms
for the new service such as ‘Acute
Liaison’ or ‘ALPS’ to monitor and
measure the frequency of the service
being referred to in the ED. Results
showed that the new service was
being referred to twice as often as the
previous service, which enabled the
Trust to feed back on the success of
the implementation of this new team.
Dashboard showing alcohol consumption trends by postcode
Lessons learned
Big Benefits Realised
Iain MacBrairdy, Business Manager of
Urgent Care at Leeds, reflects on the
NLP project: “One of the challenges is
that NLP is labour intensive, it takes a
lot of human hours to ‘teach’ the NLP,
but it’s absolutely worth it. Our use of
acronyms is probably very different
from our neighbouring hospitals, so it’s
not as simple as replicating the NLP
language across another ED. It’s
important that hospitals who adopt
NLP spend time with the team at
Ascribe teaching the system how to
interrogate the data.
From the three case studies explored,
it was clear that the ED is already
gaining real clinical benefits from
understanding more about their
service and being able to act upon the
information. The fact that data could
be used from previous years also
provides a real benefit to the ED.
Information Governance was also a
huge challenge in allowing us to use
the data for NLP – however it can be
done and it is fantastic that Leeds has
been able to set the precedence for
other trusts who want to implement
the solution.”
The NLP tool is fantastic. It allows
clinicians to own their own data,
to become more selfserving
for their own data needs for audit,
for checking out the ‘hunches’
they have and then being able to
formulate better requests to the
analysts in the information
Tiffany Watston-Koszel,
Information Analyst
Iain MacBrairdy commented “One of
the real advantages of NLP is the
ability to search on historical
unstructured data; NLP took the last
seven years of data and interrogated
that. Rather than having to change
everybody’s practise, make significant
changes to Symphony and ask all
staff to start recording alcohol in the
system, we can take the historical
trends and continue to measure the
prevalence of keywords in free text
fields, which will enable the trust to
know the impact of any service
changes made.
Another big benefit has been driving
advances in health informatics within
the trust, the project has acted as a
driver for the trust to be paper light.”
Dr. Andy Webster, Consultant in
Emergency Medicine, commented on
the new efficiencies that the ED has
gained: “We have lots of doctors and
nurses inputting data and we were
drowning in it, but not getting
anything meaningful - now with the
use of NLP, in the future we will be
able to get data in less than a few
clicks of a button. This will free up
more time for clinical resource as we
will get information much quicker.
Ascribe have been really good to work
with, they’ve been happy to
accommodate our requests for
information and have been so
enthusiastic about wanting to deliver
this project for the trust.”
Improving Patient Care
Iain MacBrairdy commented “Big Data
and NLP has delivered many benefits
to the ED, but the most important
thing is improving patient care.
Everything this project has been
delivered, from redesigning our
services to deliver more accurate
information, to accurate billing and
making sure that we’re recouping all
the income that we should, so we can
reinvest it into our services; to then
measuring the impact and the
outcomes of the things we do, is all
helping to ensure that we improve
patient care, safety and experience.”
Analyst reviewing the NLP Dashboard
Our thanks to Iain MacBrairdy,
Business Manager, Urgent Care and
Dr Andy Webster, Consultant in
Emergency Medicine for their
contributions towards the case study.
Presentation: Emergency Care:
The Journey to Big Data
Presented by Iain MacBrairdy and
Dr Andy Webster, presented at the
Ascribe Conference 2013.
YouTube: Leeds Teaching Hospitals
implements Ascribe Big Data NLP
Copyright 2014 - Ascribe Ltd.
ED staff
Management Executives
Improved patient safety.
Improved patient safety.
Improved patient safety.
Easier control and analysis of
Enables them to be more self
serving and own their data.
Cost-savings through reduction of
Reduction in handwritten errors.
Reduction of errors from
handwritten notes.
Cost-savings through improved
operational efficiencies.
Continue to use the system as they
should as Big Data/NLP can
interrogate unstructured data.
Improved report generation and
analysis of costs.
Can be integrated into the systems
the staff already use.
More time in the ED and less time
searching for information on paper
­Ascribe Ltd.
Ascribe House
Brancker Street
Bolton BL5 3JD
Provides evidence and back up for
proposals for service redesign.
Telephone +44 (0)8700 53 45 45
+44 (0)8700 53 37 77
[email protected]
Contributes towards meeting
paperless goals for NHS.
Ascribe BDEM Jnauary 2014