To determine if brain-responsive neurostimulation (RNS)-detected events had to reach a certain duration in order to evolve into a seizure and to determine if the frequency and duration of events detected by the RNS could serve as marker of clinical seizure activity.

The relationship between duration of detected RNS events and scalp EEG seizures is unknown.

Retrospective study of epilepsy patients with RNS device who underwent video EEG (vEEG) monitoring in our epilepsy monitoring unit. Patient Data Monitoring System (PDMS) was queried to analyze ECOG library tracings to quantify the number of stored long-term episodes during the scalp vEEG monitoring period. The RNS detected events were then compared to scalp vEEG recorded seizures and BIRDs. Mann-Whitney test was used to assess group differences in numeric variables.

11 patients with a mean age of 35.53 years (range: 18-59) were identified. There was a greater number of long episodes captured in the RNS ECOG library when compared to scalp EEG detected seizures (mean 10.44 vs. 1.81, p = 0.01) during the admission period (mean 2.75 days, range: 1-5). When comparing the events seen in both the ECOG library and scalp EEG to the events seen in the ECOG library only, we found that the former had a longer duration (mean 27.4s  vs. 13.2s, p < 0.0001) and higher number of delivered treatments (mean 4.5 vs. 3.1, p = 0.02). Events in the ECOG library lasting more than five seconds did not differ from the number of seizures seen in scalp EEG (P>0.05).

RNS detected events lasting longer than 5 seconds had good correlation with scalp-vEEG detected clinical (scalp EEG negative) or electroclinical seizures. This study suggests longer (> 5 seconds) lasting RNS events could be used as a biomarker to determine individual patient’s seizure frequency, and to monitor seizure frequency over time.