Blog Home

Epilet is a tinyML-powered bracelet for detecting epileptic seizures

Arduino TeamJune 8th, 2021

Epilepsy can be a very terrifying and dangerous condition, as sufferers often experience seizures that can result in a lack of motor control and even consciousness, which is why one team of developers wanted to do something about it. They came up with a simple yet clever way to detect when someone is having a convulsive seizure and then send out an alert to a trusted person. The aptly named Epilet (Epilepsy + bracelet) system uses a Nano 33 BLE Sense along with its onboard accelerometer to continually read data and infer if the sensor is picking up unusual activity. 

The Epilet was configured to leverage machine learning for seizure detection, trained using data captured from its accelerometer within Edge Impulse’s Studio. The team collected 30 samples each of both normal, everyday activities and seizures. From this, they trained a model that is able to correctly classify a seizure 97.8% of the time.

In addition to the physical device itself is an accompanying mobile app that handles the communication. When it receives seizure activity that lasts for at least 10 seconds from the Nano 33 BLE Sense, the app sends an SMS message to a contact of the user’s choice. The Epilet has a lot of potential to help people suffering from epilepsy, and it will be exciting to see what other features get added to it in the future.