Seizure forecasting: Where do we stand?

Area of Science:

• Neuroscience
• Biomedical Engineering
• Data Science

Background:

• Seizure forecasting has progressed significantly due to technological and analytical advancements.
• Multimodal monitoring using wearable and implantable devices provides ultralong-term data.

Purpose of the Study:

• To review milestones in seizure forecasting discussed at ICTALS 2022.
• To highlight the impact of multimodal data and network science on understanding seizure dynamics.

Main Methods:

• Utilizing data from wearable and implantable devices (EEG, motor, acoustic, autonomic signals).
• Applying network science approaches to analyze epilepsy as a large-scale network disorder.
• Developing and evaluating seizure forecasting algorithms with probabilistic risk assessment.

Main Results:

• Confirmed cyclical nature of interictal epileptiform activity, seizure risk, and seizures across daily, multi-day, and yearly timescales.
• Revealed novel insights into pre-ictal dynamics using network science perspectives.
• Shifted focus from discrete predictions to continuous probabilistic seizure risk forecasts.

Conclusions:

• Multimodal monitoring and network science are key drivers in advancing seizure forecasting.
• Accurate performance evaluation requires comparison with constrained stochastic null models.
• Effective communication of seizure forecasts is essential for socioeconomic impact and patient care.