This post marks the beginning of a research project into seizure prediction.

I have generalised epilepsy, after two tonic-clonic fits I experienced in December 2017. They were both preceded by an aura, which I ignored, and continued on my commute. Ultimately, I ended up collapsing, having a seizure, and being taken to hospital.

I still have seizures, but these are mostly focal/absences/myoclonic. I’m worried when I leave the house on my own, and I find it difficult.

I think the main thing is not knowing when a seizure will happen, and I’ve been shaken when the paramedics turn up, and I can’t see, and they’re taking bloods and inserting an cannula.

My other worry is that someone will take my belongings. Of course, people are more likely to help, but there are always exceptions to the rule; this is inevitable.

The thing about my seizures is that it can happen at any time, and I do live in fear of it happening. I don’t think my epilepsy is fully under control yet, either.

Recently, I’ve been beginning to think about ‘seizure prediction’. There’s been research on seizure prediction since the advent of digital EEG scans, but a large proportion of this research appears to rely on EEG monitoring.

EEG monitoring is something I’ve ruled out. Whilst its excellent for monitoring brain activity, its also uncomfortable, impractical, and certainly not portable for day-to-day long-term use.

Some approaches incoporate GSR (galvanic skin response), which is what my approach will center around, amongst other inputs.

The basis of what I’ve been researching is derived from seizure dogs. My question is the following: “If dogs can sense/smell oncoming seizures, why can’t a computer?”.

Using GSR to detect oncoming seizures, amongst other sensors, and combined with machine learning, we can continously improve detection rates, and reduce false-positives.

The end result of this project will be a wrist-worn device that interprets heart rates, GSR inputs, and accelerometer data.

To have such a device would certainly help those with epilespy to get to a safe place/position before a seizure starts.

I don’t intend to make the design proprietary; it should be open hardware, as many epilepsy devices are proprietary, and I think that such a device should be available to all.

More on this later on…