How our Seizure Detection Works

What are seizures?

Seizures are medical events characterized by excessive neuronal activity in the brain, causing visible symptoms such as uncontrollable jerking, loss of consciousness etc. Most (around 60%) seizures are convulsive in nature and can last from a few seconds up to 5 minutes.

How does convulsive seizure detection using SeizureSync work?

SeizureSync detects the shaking motion that occurs in the extremities using the accelerometer a smart watch (Pebble watch for now) as a sensor. It utilizes numerical techniques such as frequency analysis, line of length/variance changes, etc, to classify events as seizures. The time frame for seizure detection used is 5 seconds (the convulsive activity has to occur for 5 seconds for the detection to be made).

In order to make sure the algorithm detects seizures, testing was done on a seizure simulator, which used EEG data fed into a physiological control system representing an arm and the output fed into a servo motor ‘arm’. The seizure detection device is attached to the arm and checked for sensitivity (true positives/true positives+false negative) and specificity (true negatives/true negatives+false positives).

What about false alarms?

If seizure is detected, it sends an alarm within 10 seconds, giving you a chance to cancel in case of false alarms. The false alarm rate in the current algorithm is less than one a day, which is a vast improvement over past versions and other seizure detection algorithms.

What devices will SeizureSync work on?

The goal is to make SeizureSync available on all popular smartwatches so that our users can have seizure detection on the devices that they already own and love. 

Currently, SeizureSync is available on Pebble watches.

I have absence or myoclonic seizures and I want to be able to track these. Is there a way to do that?

We have made some progress with myoclonic seizure detection, but the algorithm is not production ready. With the data we will be able to collect in the near future, we hope to make significant progress in these areas that affect about 40% of epileptics.

In addition, we are also working on fall detection algorithms that will give us information that an emergency event has occurred and notify your loved ones.

Please download our Android and iOS apps, and give us feedback so that we can create a comprehensive management tool for epilepsy.