Bigger data, better healthcare
13 October 2016
As data volumes go on growing at an astonishing rate, so does the potential to put them to beneficial, life-changing use – and nowhere more than the vital field of healthcare provision.
Research IT Services (RITS) is currently applying its expertise to the pioneering NIHR Health Informatics Collaborative (NHIC), turning fast-moving mountains of sensitive patient data generated by intensive care units into a usable resource that can inform medical research, improve patient outcomes and help cut healthcare costs.
Working in support of the co-ordinating team at UCL/UCL Hospitals – a key player in the Collaborative – RITS is proving that, in this breathless age of big data, the challenge of taming, tidying and transforming datasets into real-world value can be tackled with striking success.
Harnessing a wealth of health data
21st century hospitals are data-rich environments. From regular updates on heart rate, blood pressure and drug treatment, for instance, to basic details such as gender, height, weight and NHS number, extensive information on patients is routinely collected and stored. But what if all this could be aggregated and connected more effectively, within and across hospitals, and used to help shape medical practice and enhance patient care?
That’s the aim underpinning the National Institute of Health Research’s (NIHR’s) HIC initiative. Involving five leading university hospital trusts, it comprises five individual projects focusing on high-priority fields of healthcare: acute coronary conditions: ovarian cancer; hepatitis; kidney transplants; and critical care.
The last of these is being spearheaded by UCL Hospital’s critical care team, which each year delivers top-quality care for over 2000 patients suffering from life-threatening illnesses and injuries. “We’re automating the collation of wide-ranging, longitudinal data and providing a unified point of access to it”, says Dr Steve Harris, who’s leading the project. “The result will be a big increase in the quality, availability and usefulness of electronic patient records.”
Fundamentally, the project has posed a classic big data challenge: bringing order to an ever-swelling ocean of information and turning it into actionable insights. It means coping with the sheer volume and speed of data generated by the participating units and with complex datasets differing widely in quality and format, as well as the need to:
- Produce a single, coherent resource that’s easy to interrogate, manipulate and mine.
- Maintain complete confidentiality while ensuring data remains linkable in order to understand patient pathways before and after their illness.
A critical role for RITS
The critical care units contributing to the project generate around 3000 data items for each patient they admit. Forging these into a readily usable, research-friendly resource clearly demanded specialist IT expertise beyond the team’s in-house capabilities.
“We engaged RITS to work closely with us and deliver what we needed,” Steve Harris says. “They did more than simply write software code – they developed a real understanding of the project, its purpose and its clinical dimension.”
Supported by his colleagues at RITS, Research Software Engineer Sinan Shi set about developing a software package to convert standard, unusable XML files into a uniform, queryable database. “There were 41 million data points to accommodate, covering just over 16,000 patient admissions from 2014 to 2016,” he explains. “In all, we wrote thousands of lines of code to provide the range of functions that the database needed to incorporate. As with all RITS projects, the software was developed in a sustainable fashion, with all code readable, reusable, properly tested and peer-reviewed to ensure quality.”
The outcome successfully met three key criteria:
- Clean – because the data originates from multiple sources, it has had to be deduplicated, checked, validated and where necessary corrected, e.g. to ensure heart rate and other data is in the required range, all information has been entered in the right place, the height/weight of every patient is presented in the same units etc.
- Convenient – the system has been designed so that non-IT experts with no programming skills can easily access, extract and manipulate the data they need; this will encourage them to explore and experiment, facilitating the generation of fresh insights from the data.
- Confidential – because the data contains personal details and medical histories, the need to keep it hack-proof is absolutely paramount. Identifiable data is therefore deposited in the UCL Data Safe Haven, a custom-made secure facility for data storage, handling and analysis, while an anonymised version can be safely shared among researchers.
“Without RITS, we’d have had to engage external IT consultants,” comments Dr Harris. “That wouldn’t simply have been more expensive – it would have meant missing out on the convenience of tapping into first-class expertise right on our doorstep and the extra dimension it’s added in ensuring our project benefits from a strong working bond.”
Intensive effort rewarded
Developing fit-for-purpose infrastructure for the project, though, was just the first step. “The next stage is to build tools and applications that can unlock the full value of the data,” Sinan Shi explains. “The aim is to let users unleash their imaginations and maximise the information’s potential, running reports for themselves and providing feedback for clinicians involved in hands-on intensive care.”
“Submitting and storing data is one thing, but our real goal is to make sure people exploit it on a regular basis”, Steve Harris concludes. “To that end, enabling linkages to be made across the data is absolutely crucial and we’ll be working with Sinan over the coming months to optimise that.”
The conclusion is clear. Big data really can be good for your health – and engaging the right IT expertise can be a sure cure for any doubt about its potential benefits.