Epilepsy as a dynamic disease: A Bayesian model for differentiating seizure risk from natural variability

Area of Science:

• Neurology
• Biostatistics
• Data Science

Background:

• Epilepsy treatment is challenged by seizure frequency changes that may not reflect true risk alterations.
• Natural variations in seizure occurrence can lead to unpredictability and suboptimal medication adjustments.
• Existing methods lack rigorous statistical approaches to distinguish variability from genuine risk shifts.

Purpose of the Study:

• To develop a novel statistical tool to accurately assess individual seizure risk in epilepsy.
• To differentiate between natural fluctuations in seizure frequency and actual changes in underlying seizure risk.
• To evaluate the tool's implications for understanding the natural history of epilepsy, specifically in tuberous sclerosis complex (TSC).

Main Methods:

• Developed the Epilepsy Seizure Assessment Tool (EpiSAT), a Bayesian mixed-effects hidden Markov model for zero-inflated count data.
• Utilized a large dataset from SeizureTracker.com, including over 1.2 million patient-reported seizures.
• Validated EpiSAT's accuracy through simulation and applied it to analyze TSC data.

Main Results:

• EpiSAT significantly improved seizure risk assessment compared to traditional methods.
• Identified four distinct underlying seizure risk states in individuals with TSC.
• Estimated the expected duration of each seizure risk state to be less than 12 months.

Conclusions:

• Proposed a novel Bayesian statistical approach for individual-level seizure risk evaluation using patient-reported diaries.
• EpiSAT can incorporate clinical variables to assess their impact on seizure risk.
• This tool has the potential to enhance clinical practice by distinguishing true risk changes from natural variations, improving antiepileptic drug management and treatment timing.