4 ms·

The phrasing might sound bad from a math perspective, but they're trying to convey something important about the nature of seizures. Seizures act more like a le

by sdwr 5y ago

The phrasing might sound bad from a math perspective, but they're trying to convey something important about the nature of seizures. Seizures act more like a learned skill than a typical illness. The first one primes the brain, and repeated seizures reinforce the pattern, making it easier to trigger.

harry8 5y ago
Not just might sound bad. It sounds like utter garbage in support of a theory rather than evidence. Bad mathematical reasoning here really does negate the the claim as anything more than "I wonder if ..." Could be true. Let's find out. Turning it around the probability that you have one and only one seizure could be exactly the same as the probability that you have a first seizure and it is NOT epilepsy. Those could well be different ways of saying precisely the same thing.
whatshisface 5y ago
Model A: First seizure causes subsequent seizures. Model B: Factor X causes many seizures in a sequence that, as sequences usually do, begins with the first. Any reason they prefer A over B? Maybe there is some evidence of learning, or a statistical fact that is true of seizures causing seizures but not of factor X causing them independently?
- drdeca 5y ago
  What I'm about to say is just hypothetical, I know basically nothing about medicine. Suppose that seizures at one point didn't have an influence on the probability of seizures later. Then, that seems to me like it would mean that seizures were a kind of independent point process? Like, suppose there's a model where, depending on a person's genetics, age, and, idk, let's say diet, predicts a given rate of seizures, and also that this rate varies continuously with age. In such a model, I don't think we would tend to see something where a person has no seizures until a certain point in their life, and then subsequently begins to have them at roughly a given rate for some period of time, but it would seem compatible with the "seizures causing seizures" idea. Note: I am not claiming that this is what the data actually shows. I have no idea. But, on the other hand, if there were some other invisible property which can at some random point change from state A to state B, and in state B seizures are more common, but seizures don't influence whether in state A or in state B, I'm not sure you could distinguish that from the "seizures cause more seizures" idea? But, if the causal influence wasn't just "have they ever had a seizure", but instead all seizures had an impact on chance of seizure in the near future, then I think that would have statistical effects? tl;dr: I don't know what I'm talking about, but I suspect you could at least in principle distinguish the two models using some statistical properties.
andai 5y ago
I hear similar things about psychosis, it changes the brain in a way to increase the risk of future psychosis.
elric 5y ago
This is an interesting point. It's pretty common for people to start getting migraines after a head injury (concussion etc). Apparently the genetic component might already be present, but the disease can be dormant until "activated". Perhaps this, too, could be "unlearnt" somehow.

by sdwr 5y ago

harry8 5y ago

Not just might sound bad. It sounds like utter garbage in support of a theory rather than evidence. Bad mathematical reasoning here really does negate the the claim as anything more than "I wonder if ..." Could be true. Let's find out. Turning it around the probability that you have one and only one seizure could be exactly the same as the probability that you have a first seizure and it is NOT epilepsy. Those could well be different ways of saying precisely the same thing.

whatshisface 5y ago

Model A: First seizure causes subsequent seizures. Model B: Factor X causes many seizures in a sequence that, as sequences usually do, begins with the first. Any reason they prefer A over B? Maybe there is some evidence of learning, or a statistical fact that is true of seizures causing seizures but not of factor X causing them independently?

drdeca 5y ago
What I'm about to say is just hypothetical, I know basically nothing about medicine. Suppose that seizures at one point didn't have an influence on the probability of seizures later. Then, that seems to me like it would mean that seizures were a kind of independent point process? Like, suppose there's a model where, depending on a person's genetics, age, and, idk, let's say diet, predicts a given rate of seizures, and also that this rate varies continuously with age. In such a model, I don't think we would tend to see something where a person has no seizures until a certain point in their life, and then subsequently begins to have them at roughly a given rate for some period of time, but it would seem compatible with the "seizures causing seizures" idea. Note: I am not claiming that this is what the data actually shows. I have no idea. But, on the other hand, if there were some other invisible property which can at some random point change from state A to state B, and in state B seizures are more common, but seizures don't influence whether in state A or in state B, I'm not sure you could distinguish that from the "seizures cause more seizures" idea? But, if the causal influence wasn't just "have they ever had a seizure", but instead all seizures had an impact on chance of seizure in the near future, then I think that would have statistical effects? tl;dr: I don't know what I'm talking about, but I suspect you could at least in principle distinguish the two models using some statistical properties.

andai 5y ago

I hear similar things about psychosis, it changes the brain in a way to increase the risk of future psychosis.

elric 5y ago

This is an interesting point. It's pretty common for people to start getting migraines after a head injury (concussion etc). Apparently the genetic component might already be present, but the disease can be dormant until "activated". Perhaps this, too, could be "unlearnt" somehow.