Seizure self-prediction: Myth or missed opportunity?

Published: October 03 2017
Last updated: September 29 2022

Introduction from Dr Markus Reuber, editor-in-chief of Seizure

Seizure 51 has published Seizure self-prediction: Myth or missed opportunity?

The unpredictable and intrusive nature of epileptic seizures is an important cause of distress and disability (1). It also has major effects on medical management: although the key manifestations of epilepsy are paroxysmal, the first line treatment of epilepsy, anticonvulsive medication, is almost invariably used continuously – with all the unwanted effects this approach entails. If individuals with epilepsy could predict their seizures – either on the basis of symptoms or using devices – they could limit the social effects of epilepsy, reduce the risk of seizure-associated injuries and treatment-related side effects by using medication only when needed. This makes the predictability of epileptic seizures a research topic of high importance.

The existence of seizure prodromes – physiological, cognitive or mood changes preceding seizures by hours or days, has been suspected for a very long time, and recent neurophysiological evidence suggests that the development of epileptic seizures in the brain is often not as sudden as the seizures may appear to onlookers or even those experiencing them. This work suggests that there may well be an objective basis to pre-ictal changes which people with epilepsy may perceive before a seizure or which their friends and family may notice.

Unfortunately, research in this important area is challenging. Some patients reporting warning symptoms before their seizures may only recognise these symptoms retrospectively, after the seizure is over. While it is possible that their reports are accurate, and that the only problems with the prospective use of these symptoms as seizure warnings is a lack of specificity or attention, it may also be that their retrospective reports are inaccurate, and driven by a natural urge to find explanations and specific causes for things for which these explanations do not exist (2). On the other hand, there may well be individuals with epilepsy who do not report any seizure warning symptoms although they actually experience changes, which could reliably predict seizures. People living with a disorder such as epilepsy may, conceivably, train themselves to suppress or ignore slight “niggles” to try and extend the time in their lives when they can pursue their interests and activities. Situations in which friends and family report that they are able to predict seizures on the basis of clear behavioural changes hours or even days before a seizure while the person with epilepsy themselves can’t certainly suggest that such scenarios exist.

My editor’s choice from the current issue of Seizure by Michael MacKay and colleagues is a narrative review based on a systematic search of the literature examining the area of self-prediction of epileptic seizures (3). They find that a significant proportion of patients, between 17% and 41%, are able to recognise periods of increased seizure risk based on a wide variety of easily measurable factors such as mood, sleep, and prodromal symptoms. Their review also suggests that the design of previous studies may have underestimated the proportion of patients capable of doing this. However, they also find that these symptoms can precede a seizure by up to 12 hours, and that it has not proven possible to identify specific symptoms which reliably predict the imminent occurrence of a seizure. Hence, whilst some patients can foresee a seizure within the next 12 hours or so at levels significantly better than chance, the precise time at which this will occur usually remains impossible for them to predict.

Despite these limitations the authors conclude that this predictive ability could provide a basis for better-targeted pharmacotherapy. In addition correlations
between self-perceived seizure risk and additional self-reportable factors such as anxiety, stress, sleep and cognitive prodromal symptoms could provide more accurate seizure predictions if they were combined in advanced calculations, for instance involving machine learning algorithms which could be incorporated in mobile devices. Portable devices could use additional physiological measures such as body temperature, heart rate variability or skin conductance changes to determine the risk of seizures. Conceivably, such methods could also be used to train individual patients to become better observers of their pre-ictal prodromes and predictors of their seizures.

1) Novakova, B., Harris, P. R., Ponnusamy, A., Reuber, M. The role of stress as a trigger for epileptic seizures: a narrative review of evidence from human and animal studies. Epilepsia. 2013;54(11),1866-76.

2) Sperling et al. Self-perception of seizure precipitants and their relationship to anxiety level, depression and health locus of control in epilepsy. Seizure 2008;17:302-307.

3) Mackay M, Mahlaba H, Gavillet E, Whittaker RG. Seizure self-prediction; myth or missed opportunity? Seizure 2017; 51