Detection and differentiation of epileptic seizures using wearable sensors in smart textile

Important results from the project

The aim was to investigate whether our sensor garment can detect and differentiate different epileptic seizures. The goal was to develop machine learning algorithms with high sensitivity and specificity that can distinguish between different types of epileptic seizures and also differentiate from non-epileptic seizures. Accurate recording of seizure type and frequency would help optimize treatment of epilepsy. Due to the pandemic, we have not been able to test the garment during seizure registration for long periods of time. Furthermore, one seizure type was too infrequent.

Expected long term effects

We have recorded patient data with the sensor garment from 44 patients during the project period, a total of 182 recorded seizures, and further developed an algorithm to detect tonic-clonic seizures. We had to stop trying to create algorithms for hypermotor seizures. We have shown that there are now prerequisites for the detection of psychogenic non-epileptic seizures in order to differentiate between them and tonic-clonic epileptic seizures.

Approach and implementation

The plan was to use a previously developed garment with built-in sensors for measuring several physiological variables to collect data on epilepsy patients under investigation to investigate whether the shirt can detect and differentiate epileptic seizures and a type of non-epileptic seizure. The goal was to develop machine learning algorithms with high sensitivity and specificity that can distinguish between different types of seizures. Due to the low number of registered attacks, we could not fully meet the objective