A new study from Italy has shown that a computer model using electroencephalogram (EEG) data can predict whether a person will respond to the epilepsy medicine levetiracetam.
The researchers, Pierpaolo Croce and colleagues, created a machine learning model based on data from EEG tests, which are used to record brain activity in people with epilepsy. Machine learning models are computer programs that process data and can use that to create predictions for new data.
The group used EEG data from 23 people with temporal lobe epilepsy to test whether it can predict how well levetiracetam would work.
EEGs were done before people started to take levetiracetam and three months after starting. After two years of the study participants taking the medicine, the researchers grouped them into those who were seizure free and those who were not seizure free.
The team found 152 features in the EEG data that could help predict whether a person’s medicine would work for them or not. Their machine model showed that it could predict the effectiveness of levetiracetam in about three quarters of cases when using just data from before the medicine was started. When using data from before the medicine was taken and data from three months after starting, the model could predict correctly in about four in five cases.
The study, published in the journal Clinical Neurophysiology, concluded that this shows the possibilities of machine learning models using EEG data in predicting medicine effectiveness in people with epilepsy. The researchers added that future studies should use this to try to develop a model that can match people with epilepsy to the epilepsy medicine most likely to work for them.
You may also be interested in...
New technology to make magnetoencephalography (MEG) brain imaging more accessible in children has been developed by UK scientists and Young Epilepsy.