Dude

[WilliamBelini]

Hello.
Sorry about my ignorance. Can I utilize this algorithm to analyze realtime data from EEG, and identify in the multiple chanels, what them had more corelation with "future" seizures? I imagine it can clusterising pre-seizures signals based in time and spectrum variance for different type of epilepsy, utilizing a chanel to design the seizures and utilizing others to identify possible pre-seizures signals. I can't found a product wearable to predict epilepsy seizures. So I'm finding some EEG wearable and IA algorithms to make a realtime advisor to help my mother.

[truongduynhan]

Hi,
Yes, you certainly can use it as a realtime predictor of seizures. Time needed for pre-processing and AI inference is far less than 30s which is the length of each EEG window being fed into the algorithm. Performance varies from one person to another, but you have great advantage of collecting very long-term EEG from the same person. That would help improve the accuracy of the AI.