Systematic comparison of Commercial seizure detection Software: Update equals Upgrade?

To compare two updated commercially available seizure detection software packages (encevis 2.0 and Persyst 14) in a large video-EEG monitoring dataset comprising 80 patients with 6826 h of EEG and 912 seizures.

We conducted a follow up study of two seizure detection software packages with a significant update since 2021 using different sensitivity settings.

Mean overall seizure count was 11.4 and mean tonic-clonic seizure rate was 0.7 per patient. Detection rates (DR) per patient were 80.7% to 89.1% for encevis 2.0 and 67.3% to 71.1% for Persyst 14. False positive rates (FPR) per day were 12 to 22 for encevis 2.0 and 2 to 4 for Persyst 14.

Encevis 2.0 showed a significantly higher DR than Persyst 14, whereas Persyst 14 showed a significantly lower FPR than encevis 2.0.

Updated FDA and CE certified seizure detection software showed opposite changes: encevis improved in sensitivity and deteriorated in FPR, while Persyst improved in FPR and deteriorated in sensitivity. In agreement with previous publications, automatic seizure detection software is ready for routine clinical use, but specialized personnel with sound neurophysiological knowledge and awareness of the strengths and weaknesses of the specific software package used is critical.