82% accurate epilepsy detection method developed via crowdsourcing site
A crowdsourcing site has used the power of people and the internet to help launch a new method to detect, predict and prevent epileptic seizures.
Crowdsourcing, not to be confused with crowdfunding, is the online process of gaining help, expertise, service, ideas or content from the online community.
A competition launched by the online crowdsourcing platform ‘Kaggle’, which calls itself ‘The Home of Data Science’, to help develop a method to detect, predict and prevent epileptic seizures.
The competition, hosted by Kaggle, invited people from all over the world with access to the internet to develop a method to help. The crowdsourcing site received entries to the competition from huge numbers of people across the world, with a huge diversity of backgrounds in everything from mathematics to computer science and engineering.Currently, the treatments for epilepsy range from medication to surgery and implanted electrical device which use electrical pulses to help prevent the onset of a seizure. Until now, doctors and healthcare professionals around the world had been at a loss when it came to ‘fine-tuning’ treatments so that they would be effective when a seizure strikes a patient.
The competition was sponsored by the American Epilepsy Society and the National Institutes of Health’s National Institute of Neurological Disorders and Stroke and the Epilepsy Foundation. Instead of adopting the traditional approach of asking a handful of research labs to look into developing a method of detection and prevention of a seizure, they uploaded huge amounts of data recorded from the brains of dogs and humans as they had seizures over a number of months. They then challenged people all over the world to use this information to develop an accurate and unique method.
The entrants were set two challenges – one to find a method of seizure detection and another for seizure prediction.
The winning ‘prediction’ team comprised of five participants from the United States and Australia, and featured a software engineer and a mathematician – but no doctor. They managed to compile data on the electrical activity in the brain to develop an algorithm which forecasts epileptic seizures an impressive 82% of the time.
The detection contest was won by a software engineer, Michael Hills. A good portion of patients with epilepsy do not always respond to medication, so Hills stated that their best hope is ‘responsive neurostimulation devices’ which detect seizures as quickly as possible and activate to stop them.
The competitions sponsors hope that these results will soon be used in conjunction with devices to be used by epileptic patients to help their condition.