Automated real-time detection of tonic-clonic seizures using a wearable EMG device

Abstract

ObjectiveTo determine the accuracy of automated detection of generalized tonic-clonic seizures (GTCS) using a wearable surface EMG device.MethodsWe prospectively tested the technical performance and diagnostic accuracy of real-time seizure detection using a wearable surface EMG device. The seizure detection algorithm and the cutoff values were prespecified. A total of 71 patients, referred to long-term video-EEG monitoring, on suspicion of GTCS, were recruited in 3 centers. Seizure detection was real-time and fully automated. The reference standard was the evaluation of video-EEG recordings by trained experts, who were blinded to data from the device. Reading the seizure logs from the device was done blinded to all other data.ResultsThe mean recording time per patient was 53.18 hours. Total recording time was 3735.5 hours, and device deficiency time was 193 hours (4.9% of the total time the device was turned on). No adverse events occurred. The sensitivity of the wearable device was 93.8% (30 out of 32 GTCS were detected). Median seizure detection latency was 9 seconds (range −4 to 48 seconds). False alarm rate was 0.67/d.ConclusionsThe performance of the wearable EMG device fulfilled the requirements of patients: it detected GTCS with a sensitivity exceeding 90% and detection latency within 30 seconds.Classification of evidenceThis study provides Class II evidence that for people with a history of GTCS, a wearable EMG device accurately detects GTCS (sensitivity 93.8%, false alarm rate 0.67/d).

Tonic-clonic seizure detection using accelerometry-based wearable sensors: A prospective, video-EEG controlled study

“…Consequently, in comparison with other studies, our results achieved comparable sensitivity and FP rate using only one modality, without inclusion bias and without any movement restrictions in the algorithm development and evaluation. A variation in the number of FPs was noted between individuals in relation to heterogeneous motor phenomena of TCSs in this study, similar to other seizure detection devices [20][21][22]. Two individuals (ID 59 and 93) with generalized onset TCSs generated a higher false positive rate in the KNN algorithm than in the other two algorithms, and generally high false positive rates when compared to patients with focal onset TCSs.…”

Standards for testing and clinical validation of seizure detection devices

“…These studies qualify for phase 2. Larger multicenter studies (69‐199 patients) used dedicated wearable devices (sEMG and accelerometer combined with electrodermal activity, respectively) . However, data analysis was offline (not real time) from archived data, and the algorithm was tested retrospectively, across a wide range (95‐255) of threshold‐values, or optimized using the leave‐one‐out method .…”

Detection of convulsive seizures using surface electromyography

“…Therefore, when the device was placed more than 45 degrees from the midline of the belly of the biceps (this was the case in 29 patients included into the study), crosstalk between biceps and triceps muscles caused a reduction in the signal amplitude (in‐phase cancellation), greatly reducing sensitivity of the algorithm . In the EDDI system, electrodes were placed parallel to the muscle fibers, which prevented crosstalk and in‐phase cancellation . In a subgroup of patients in whom the device was placed over the midline of the biceps muscle (n = 149), the sensitivity reached 100%: the device detected 29 of 29 TCS (95% CI 88%‐100%) that occurred in the 24 patients with a properly placed device, with a latency of 7.7 s and FAR of 1.44/24 h .…”

“…27 In the EDDI system, electrodes were placed parallel to the muscle fibers, which prevented crosstalk and in-phase cancellation. 26 In a subgroup of patients in whom the device was placed over the midline of the biceps muscle (n = 149), the sensitivity reached 100%: the device detected 29 of 29 TCS (95% CI 88%-100%) that occurred in the 24 patients with a properly placed device, with a latency of 7.7 s and FAR of 1.44/ 24 h. 27 Mild to moderate adverse events were reported in 28%, and 15% of the recruited patients withdrew from the study during EMG monitoring. 27 An important feature of the SPEAC system is that EMG data are stored for offline review by practitioners.…”