Novel insights into the dynamics of intractable human epilepsy

TL;DR

This study analyzes seizure dynamics in drug-resistant epilepsy, revealing power law distributions and temporal dependencies that suggest seizures can trigger each other and may be predictable over short time scales.

Contribution

It provides new insights into seizure timing and clustering, highlighting power law behaviors and potential predictability in intractable epilepsy.

Findings

Seizure and inter-seizure interval distributions follow power laws.

Increased seizure probability occurs 30 minutes before and after a seizure.

Seizures may trigger subsequent seizures, leading to clustering.

Abstract

Probability density functions and the probability of Sz occurrence conditional upon the time elapsed from the previous Sz were estimated using the energy and intervals of SZ in prolonged recordings from subjects with localization- related pharmaco-resistant epilepsy, undergoing surgical evaluation. Clinical and subclinical seizure E and ISI distributions are governed by power laws in subjects on reduced doses of anti-seizure drugs. There is increased probability of Sz occurrence 30 minutes before and after a seizure and the time to next seizure increases with the duration of the seizure-free interval since the last one. Also, over short time scales, ``seizures may beget seizures.'' The cumulative empirical evidence is compatible with and suggests that at least over short time scales, seizures have the inherent capacity of triggering other seizures. This may explain the tendency of seizures to cluster and evolve into status epilepticus. Power law distributions of E and ISI indicate these features lack a typical size/duration and may not be accurate criteria or sufficient for classifying paroxysmal activity as ictal or interictal. This dependency and the existence of power law distributions raise the possibility that Sz occurrence and intensity may be predictable, without specifying the likelihood of success.