Implementing a Network-Based Model of Epilepsy with Numpy and Numba | Hacker News Reader

Implementing a Network-Based Model of Epilepsy with Numpy and Numba

danielegrattarola.github.io

70 pointsphait12 comments

As an epileptic mathematician, I'd like to say thanks for making this. It was both interesting and informative.

As an extension, could I suggest looking at different types of seizure? E.g. tonic clonic vs Grand Mal

Hey, you're welcome! I will definitely try to expand this in the future, although I don't think that this particular model will be suitable. It's excessively simple for such a complex phenomenon. But there's a ton of literature out there, so sure!

Thanks for sharing! It's very intriguing work. Is each node intended to model the local field potential of a given brain region? How 'large' or what scale are the individual nodes? (e.g. a few hundred neurons, a handful of neurons, one neuron?)

Definitely macro scale.

I'm very new to this whole field, but usually to model the LFP of single neurons they use MUCH more detailed descriptions, down to individual membrane potentials.

This is probably not even a good model of LFP in general, it's more of a system that behaves like a brain that has seizures. They have a full discussion about it in the paper. The interesting thing about this is that by changing the connectivity you also change the attractors of the system, meaning that a "seizure" can happen or not depending on whether some nodes are connected.

For the overwhelming battery of testing I had before lobeectomy (EEGs, MRIs and CTs of course, but also PET, MEG and SEEG) I'd have never imagined that anybody was doing academic research from a timing perspective. I've subsequently dug into A LOT of the physiological research but this is all new to me. The math is WAY beyond me but the general concept is intriguing.

Was it a successful surgery? Which temporal lobe was it on?

Right temporal lobeectomy. Bill Binghamon resected 3 discrete parts. Amygdala, Hippocampus and temporal lobe. The post-op MRI is pretty wild. I'm living alone now and working mostly from home so I could be having seizures and not know but I haven't had anybody say anything. I was having complex partial seizures so unless I drop something and not notice it until later or suddenly realize I peed a little I could never know.

Wow, you had a lot of brain matter removed. I just got out of video EEG yesterday and am hoping I can have surgery, or that it otherwise goes away. Mine is left temporal lobe, which seems like a delicate place for them to operate on. Seems like it’s in the neocortex though, so I’d probably keep my hippocampus and amygdala. I still have a lot of testing left - I don’t know how I’m going to do it all. The video EEG was bad enough. Any regrets or defecits?

You could say that there are two main aspects to epilepsy research. One is on the medical side, surgery, drugs, etc., one is more about understanding the phenomenon by studying brain activity and how it behaves. The two are very strongly connected, of course, but the second can also be tackled by engineers, mathematicians, physicists or computer scientists.

You'd be surprised by how many different people work on this subject!

This sounds fascinating. I wasn’t aware this was an active field of study. Do you happen to know of any resources or literature that can give me an introduction to this or better help me understand the specifics on modelling epilepsy mathematically?

This is great. Davvero impressionante. Thanks!

Actually the mathematics of epilepsy are as intriguing as those of tumor propagation. With a lot of potential uses not only in understanding, but also in potential therapeutics.

Grazie :D

I agree, it's really fascinating. The thing that interests me the most are that many natural phenomena can be modeled as a network of some sort. It's mind-blowing how everything is connected.

Reply on news.ycombinator.com