Abstracts

This is a Late Breaking abstract

Rationale: A challenge of video-electroencephalography (vEEG) monitoring is capturing epileptic activity (interictal epileptiform discharges (IED) and/or seizure activity) during monitoring. The current study evaluated the utility of a mobile seizure risk app to increase the yield of vEEG by prospectively scheduling monitoring sessions during forecasted high risk periods. Seizure risk forecasts were first validated in a pseudo-prospective analysis of ambulatory vEEG studies undertaken in Australia between 2019 and 2022. A prospective cohort study to schedule vEEG was then initiated in 2021.

Methods: Validation data were extracted from a database of ambulatory vEEG monitoring studies (Seer Medical) and included adults diagnosed with active epilepsy (aged 18+) with a linked electronic seizure diary (>10 reported seizures) in the 6 months preceding the vEEG monitoring session (N=206 studies). Seizure events recorded in a mobile seizure diary were used to detect an individual’s multiday cycle and forecast seizure risk for future dates. For 25 of these studies, the vEEG session occurred during a monitoring timeframe deemed high risk (>25% of time in high risk according to the seizure risk forecast). The proportion of cases in which a high risk forecast corresponded with epileptic activity (abnormal report) relative to baseline (no forecast applied) was examined. For the prospective cohort study, 35 adults referred for vEEG monitoring with a diagnosis of active epilepsy (>1 seizure per month) were recruited. After 6 months (or ~10 seizures), a follow up vEEG monitoring session was scheduled according to their forecast of seizure risk (Figure 1). Outcome measures: (1) EEG reports (abnormal or normal report) and (2) physiological measures (wearable biomarkers and cortisol).

Results: In the validation study, during the high risk timeframe, 21/25 (84%) presented with an abnormal report and 4/25 (16%) a normal report. This is a 21.5% increase in identification of epileptic activity relative to baseline studies (62.5%). Similarly, during baseline vEEG session of the prospective cohort study, 63% of participants presented with epileptic activity (49% with IEDs but no seizures, 14% with both IEDs and confirmed seizure events), while 37% of participants had normal EEG. The study is on-going; however, to date, 11/35 participants completed their scheduled vEEG, with 75% of those with a high risk forecast presenting with epileptic activity (abnormal report). For instance, P1 presented with a normal EEG (baseline) and in the scheduled vEEG (in high risk) presented with abnormal EEG (epileptic activity) and physiological anomalies (Figure 2).

Conclusions: Data from this study suggest that timeframes of heightened seizure risk (identified by seizure forecasts) correspond with presence of epileptic activity. Moreover, our prospective study provides initial support for the efficacy of utilizing personalized seizure risk forecasts to schedule monitoring and optimize the diagnostic yield of vEEG.

Funding: BioMedTech Horizons Program (Australian Government), American Epilepsy Foundation’s "My Seizure Gauge" grant