Data, of course, has always been crucial to all types of research. As data scientist, Daniel Moran once said, “You can have data without information but you cannot have information without data”.
In epilepsy research, as in our day to day lives, the scale of the data available is changing dramatically. We are familiar with a megabyte (a million bytes) and a gigabyte (a billion bytes) as measures for data. But there’s so much data being generated in the world that recently new measurement unitshad to be invented – the ronnabyte (1 followed by 27 zeros) and quettabyte (1 followed by 30 zeros)!
This ‘Big Data’ is both part of, and one of the main areas of focus for epilepsy research. You can get a clear sense of this by browsing the Epilepsy Research UK research portfolio. ‘Big Data’ is generated along all parts of the epilepsy journey; genetic sequencing data, detailed EEG and MRI data, electronic patient seizure and symptoms diaries, smart phone and watch data and social media posts. All of this data, with suitable analysis tools such as artificial intelligence (AI), can answer the important epilepsy research questions.
Part of our research at Swansea University focuses on using routinely collected data for epilepsy research. This includes GP, hospital, and ambulance as well as social and educational data. We analyse this data anonymously within the SAIL databank, which ensures that appropriate safeguards are in place so that the data can be analysed safely. One of the biggest advantages of this type of research is that you can look at everybody who is living with epilepsy within the population – everybody who has had some kind of health contact can be included. The results are then more relevant to everybody with epilepsy.
One of the disadvantages with using standard, routinely collected data for research is that it doesn’t always contain enough detailed epilepsy information. This is understandable given the time constraints within the NHS – during a standard nine-minute GP consultation it is hard enough to record even the bare minimum information. One way around this is to automatically extract detailed information hidden in “free text”, for example, epilepsy specialist letters, electronic patient diaries or investigation reports. Our team at Swansea University Medical School are developing systems to do this using natural language processing technology. This technology enables computers to “understand” text written by humans, which can then be used for research.
I think ‘Big Data’ offers an exciting opportunity to make a difference for people living with epilepsy. It will be a key part of the day-to-day management of epilepsy, as well as research into the condition. In the not-too-distant future, I see AI systems using MRI, EEG and genetic data, alongside videos and smart phones, to help make more accurate and quicker epilepsy diagnoses. Adding this data to other routinely collected data on a population level will enable ‘machine-learning’ techniques to help predict prognosis and outcomes. This could help reduce the risks of things like heart attacks and bone disease for people with epilepsy and reduce the current ‘trial and error’ involved in anti-seizure medication selection.
‘Big Data’ could also be used effectively to look at treatment and outcome patterns nationally and internationally, hopefully producing the right data to drive a reduction in the current inequalities in epilepsy care – both in the UK and throughout the world.