There are companies that sell equipment for automated analysis of wake, sleep even rem sleep. I think one of them is Noldus (roughly 10-20,000 dollars).  

Don't know enough about the epilepsy model, but continuous video recording, with continuous monitoring and online analysis of the video is a possibility.  We do have  method for tracking animal movement, and limb movement, but whether it would work in this setting is another matter.  What you want is to have some data where you know what happens during your seizure, is it generalized, or more like an absence seizure.  

Our paper was published in eNeuro, the title has "Pixying behavior..."

Dear, dr. Sachdev,

I think your approach to analyzing behavior is very interesting and it can be useful for processing video with seizures.

 There is a really problem with processing of data in the experimental model, where animals develop spontaneous convulsions. These seizures occur at a random time. For detailed analysis, the behavior of experimental animals must be analyzed continuously for many days. The seizures are very specific patterns of behavior, which very differ from the patterns of normal behavior.

 We want to detect the fact, the quantity and the severity of seizure episodes.

I suppose that the solution of the problem of automatic seizure detection is interesting not only for us, but also for many groups dealing with the pharmacology of epilepsy.

 If  it is  interesting for you, we can provide some video for you with seizure to discuss the capabilities of your program in the resolution of this problem.

 Best regards

Article Differences in Evolution of Epileptic Seizures and Topograph...