“In practice, we have been on this journey for many years,” suggests John Crawford, healthcare leader, Europe, with IBM. “In the 1960s, we started to collect data from patients’ records, but what we are seeing today is much larger data extracts from databases but, in truth, we have been collecting records since Jon Snow started mapping cholera outbreaks in Soho in the 1850s.
“Data helps healthcare services to consider where best to apply their limited resources whilst improving patient health outcomes,” he explains.
Big Data is used extensively in medical research and by pharmaceutical companies to test new drugs, but today clinicians are mining data from medical records to assist them at the point of care to make better decisions.
“Risk calculators, for example, are used in hospitals to calculate the chance of a patient contracting a specific disease by incorporating all known risk factors, such as age, ethnic background, family history and the presence of symptoms, in order to determine the best treatment,” explains Crawford. “Data can also be used to predict future risks and outcomes. There’s a lot of historical data available that can be used to predict the risk of someone having to be admitted to hospital in the next 12 months.”
Being able to predict or anticipate future care needs and, if possible, intercede with patients before illness strikes has given rise to a new concept called Population Health Management. While it has been around for some time, it is only now that it is being taken seriously by healthcare organisations.
“Population Health Management is about managing health behaviours in a holistic and systematic way,” Crawford says. “In effect, it’s providing proactive, rather than reactive, healthcare – replacing the traditional approach where clinicians are confronted with an illness, then respond with a treatment.
“Data derived from multiple sources: local authorities, GP practices and hospitals – most of which is generated automatically – is monitored continuously. It is then used to predict future health risk and events and to identify those most at risk in the population in order to avoid future hospitalisations.”
“Wearable technology is an interesting area,” says Crawford. “We are seeing an explosion in consumer devices such as: activity trackers, pulse and even continuous blood pressure and blood glucose monitors, and the price is falling all the time. But where does it fit as a source of actionable data?”
Because most healthcare systems are obliged to keep regulated medical records that are separate from data supplied by patients, the medical profession is not in a position to consume data from wearable devices.
“A number of countries are looking to merge data from wearables with medical records; for example, Scotland and several regions in Spain are looking to combine personally generated healthcare data with clinical healthcare data,” Crawford suggests.
There are concerns, though, that by incorporating data from consumer wearables, carefully constructed medical data will become ‘tainted’ and recent legal action by ‘FitBit’ users in the US – suggesting the devices were consistently misrecording their heart rates ‘by a significant margin’ –raises issues about the quality of data provided.
“What we need is small data,” suggests Crawford. “Big Data is great for crunching large amounts of data; small data is what we need if we are to draw insightful, actionable information that will help the clinician do better for the patient.”
Sharing information is an issue as wearable devices are tethered to an application and then to a database that has not been designed for sharing.
“We have to get to a point where it will be easier to interoperate and share that data,” Crawford suggests.
Modelling of data
Much of the data generated in medical record systems is populated by doctors or administrators and comprises of written documentation and, more importantly, images.
“Both need to be processed differently,” explains Crawford, “whether that is narrative textual data which people have written or medical images or scans.”
As a result healthcare professionals are turning to a diverse set of cognitive solutions.
“Much of the data is not codified or structured,” says Crawford. “It comprises of images, text or waveforms for example. The data might not be clean, there may be a lot of noise, hence the development of IBM Watson and the growing use of cognitive computing, that can learn and develop expertise over time.”
Last year, IBM announced plans to acquire Merge Healthcare, a developer of a medical imaging management platforms and a provider of medical image handling and processing, in a move intended to provide access to and unlock the value of medical images to help physicians make better patient care decisions.
Research conducted by IBM suggests that medical images are by far the largest and fastest growing data source in the healthcare industry; IBM researchers estimate they account for at least 90% of all medical data.
The volume of medical images, much like data in general, can be overwhelming, even to the most sophisticated specialists. Tools to help clinicians extract insights from medical images remain very limited, requiring most analysis to be done manually and as a result medical images tend to remain largely disconnected from mainstream health information.
As a result of acquiring Merge, IBM Watson will be able to provide an ability to cross-reference medical images against, for example, laboratory results, e-health records, tests, clinical studies and other health-related data sources, helping to generate detailed insights to help healthcare providers in areas as diverse as radiology, cardiology, orthopaedics and ophthalmology.
“As a result, clinicians will be able to provide a more personalised approach to diagnosis, treatment and monitoring of patients,” Crawford explains.
“We are going through a period of exploration and discovery as to what we can do with big data,” Crawford suggests. “In the pharmaceutical industry, its use is well understood, well established and accepted. There is certainly the potential to do a lot more with the data in terms of the way in which the healthcare system works and to provide more personalised healthcare.
“Healthcare professionals understand the power this data can provide and want to get their hands on it. However, the medical profession works slowly and, in many cases, haven’t embraced technology in the way that other sectors have. That needs to improve if the benefits of Big Data are to be fully realised.”