Abstracts

Rationale: Monitoring motor seizure activity outside of a clinical setting is based on patient-reported outcomes. This complicates determination of the efficacy of anti-seizure medications and poses a challenge in clinical practice. To quantify differences between epileptic and non-epileptic motor recruitment, sEMG data from generalized tonic-clonic (GTC) seizures, psychogenic non-epileptic seizures (PNES), and trained simulated seizures were captured using the SPEAC System® (a wearable sEMG monitor). Methods: Single-channel sEMG data from 25 GTC seizures and three PNES were previously recorded unilaterally from the biceps brachii at 1kHz in a prospective study of the SPEAC System. Six GTC seizures were “simulated” by healthy volunteers wearing the SPEAC System, whom were trained to exert sustained muscle contraction followed by rapid intermittent contractions. Muscle activity from each group were isolated and evaluated in MATLAB by comparing: clonic burst duration and slope of the burst duration as the seizure progresses, inter-burst interval and the change in inter-burst interval as the seizure progresses, the slope of the initial discharge (first 30ms, normalized to burst rms amplitude), and the time to maximum amplitude of the clonic bursts. An example of a captured GTC seizure with an outline of the automated isolation of the clonic bursts are shown in figure 1. The isolated bursts were overlaid on each other as shown in figure 2, and all bursts for a single event were averaged in time. All statistics were evaluated using unpaired two-tailed t-tests (p<.05) with Benjamini-Hochberg correction. Results: Energy bursts from GTC seizures, PNES, and simulated seizures lasted an average of 150.3 ms, 101.51 ms, and 217.3 ms respectively, and the slope of the burst durations were -1.35 burst#/s, -0.86 burst#/s, and -0.20 burst#/s, respectively. Inter-burst intervals from GTC seizures, PNES, and simulated seizures lasted an average of 316.19 ms, 231.37 ms, and 212.35 ms respectively, with average slope of inter-burst intervals of 21.26 ms, 4.67 ms, and 16.33ms, respectively. Burst duration was longer in simulated seizures and inter-burst intervals were longer in GTC seizures but significance was only found before post-hoc corrections. Clonic bursts and simulated bursts had normalized initial discharge slopes of 0.52 µV/µV/ms, 0.48 µV/µV/ms, and 0.42 µV/µV/ms, respectively. The time to maximum amplitude for each clonic bursts, PNES bursts, and simulated bursts was 58 ms, 52.58 ms, and 67 ms, respectively. Biometrics from the initial discharge slope and maximum burst amplitude were not significantly different. Conclusions: Clonic motor activity that occurs during GTC seizures is consistent across patients and differs from non-seizure activation. Burst duration, inter-bursts intervals, and slopes of each feature as the events progresses, although not individually significant, all contribute to the ability to discern between these different events. Continuous sEMG recording in a patient’s home/community environment may be able to differentiate between GTC seizures, PNES, and non-seizure rhythmic activity. There is still a wealth of information to be mined and utilized from the sEMG data obtained from a continuous setting. Funding: No funding