Big Blue

The NHS in Wales faces many challenges; none bigger than improving the health of a population that is one of the poorest in Europe and has some of the highest rates of cancer and heart disease in the EU.

A partnership of organisations is hoping to tackle this issue with the help of an IBM supercomputer known as ‘Blue C’ – which has many similarities to the company’s most publicised supercomputer, Watson.

Focusing computing on life sciences

In 2007, Swansea University, IBM, and the Welsh Assembly Government clubbed together to provide the necessary funding for the university to open an Institute of Life Sciences, a purpose-built, high-performance computing facility.

At the heart of the HPC is Blue C, which collects and cleans data from hundreds of sources, links matching records together, and anonymises them – providing a unique resource for researchers to analyse.

David Ford, the man responsible for the centre and director of health informatics at Swansea said: “Blue C is unique in that it is one of the only computers in the world that is dedicated specifically to life science. It has the ability to extract meaningful answers from previously inaccessible data.”

Blue C hosts a data warehouse known as the Secure Anonymised Information Linkage databank. SAIL receives a range of information about health in Wales, including all hospital inpatient stays, day cases, outpatient visits and data from 168 of the Welsh GP practices.

The sheer size of data collection in the principality is the envy of many nations, including England, and SAIL is probably the most complete retrospective electronic patient history currently available anywhere in the world.

Ford told eHealth Insider: “What Wales does very well is the collection and release of data by organisations. Our colleges, the NHS, and other public sector organisations have been the best at that. SAIL allows us to mobilise data that was previously static in many different systems.”

Given its access to such a significant amount of data, the majority of which is hugely confidential, data protection is an obvious issue.

However, Ford said that the threat of an invasion of patient privacy had been addressed, with the SAIL databank being developed with a set of robust privacy protecting technologies, which overlay all of the anonymised data.

“We use extremely sophisticated privacy protecting techniques to de-identify the data and ensure that identities remain completely protected,” he stressed.

Watson wins Jeopardy!

IBM’s supercomputers gained great media attention last November when Watson beat two ‘legendary’ participants on the US hit game show, Jeopardy! To emerge victorious, Watson was required to behave in a manner that computers are not stereotypically built to do.

Listening and understanding natural language questions might seem simple to humans, but the convoluted manner in which Jeopardy! poses its questions would present a plethora of problems for any normal computer.

However, Watson processes natural language quickly and efficiently, and by using the wealth of information provided by its databank, it emerged victorious over its human competitors, Ken Jennings and Brad Rutter.

Ford said: “By adopting similar methods with Blue C, we would be able to get even greater benefits from SAIL, as the majority of data held in NHS and government systems is still unstructured. This is clearly a very exciting time for health informatics.”

Blue C opens up health data

The Welsh Assembly Government and the UK Department of Business, Innovation and Skills, have pumped £40m into the HPC project through an initiative that aims to give those involved in commercially-focused research access to the most advanced and evolving computing technology available.

It is hoped that its collection of data will provide significant information, which can then be used for the benefit of the NHS, improving both delivery of care and patient experience.

Andy Grant, IBM’s high performance computing sales lead in the UK, highlighted the possibility of the system providing insight into regional variations in the incidence of diabetes in Wales, which he said would previously have been impossible without investing in large surveys at great cost.

Swansea University is also initiating a second phase of the project, in which it will start to elicit feedback from patients about the quality of their treatment through a link up with Abertawe Bro Morgannwg University Health Board.

As part of this, it will deliver an e-portal for the local population of 500,000 people. The portal, delivered over web pages and smart phones, will be used initially to obtain feedback on Abertawe Bro Morgannwg’s plans.

But it will quickly be extended to measure patient experience, and then clinical outcomes such as quality of life. The technology is due to be formally launched in May with further in-line development over the following 12-18 months.

Next steps for supercomputing

The latest phase of the project has also seen an upgrade to Blue C with IBM’s new POWER7 architecture, including four new servers. The new hardware allows the computer to reach an impressive performance speed of four teraflops, according to the recognised Linpack benchmark.

The new architecture is also greener. The servers are three to four times as energy-efficient as the POWER6-based machines previously used by the system, which should save the university tens of thousands of pounds per year in electricity costs.

The development of Blue C, and its deployment in a health informatics environment, mirrors similar advances for Watson in the US. After its triumph on Jeopardy! the supercomputer is undertaking a ‘residency’ at a Manhattan cancer hospital.

Memorial Sloan-Kettering Cancer Center is to ‘feed’ Watson with textbooks, medical journals, and with permission, individual medical records. It is hoped that Watson will eventually provide express diagnoses and recommend specific treatments, ranking several alternatives, according to their suitability to the patient.

IBM says this holds out the possibility of treatment plans being constructed that not take account of medical factors but patient preferences. For example, supercomputers might be used to evaluate cancer treatments while taking into account that a patient feels strongly about not losing their hair.

However, Ford was keen to stress that the benefits of Swansea’s supercomputer project go beyond healthcare research.

He points out that by making great strides in technology and boosting the skills and employment opportunities for local people, Swansea’s economy should be improved considerably. Which, in and of itself, will make a contribution to improving health in Wales.