Journal: Frontiers in Neurology
Article Title: Wearable Reduced-Channel EEG System for Remote Seizure Monitoring
doi: 10.3389/fneur.2021.728484
Figure Lengend Snippet: Sensitivity vs. false detection rate. The sensitivity (%) and false detection rate (FDR—FP/h) are shown for the individual reviewers [green x (R1), square (R2), and circle (R3)], the consensus review (blue star), and the automated algorithm (black diamond). The unfilled markers denote the naive epileptologist review and the filled markers denote review with automated algorithm annotations as clinical decision support software (CDSS). Reviewers R1 (green x), R3 (green circle), and the consensus (blue star) showed improved sensitivity with CDSS at the expense of a higher FDR, while reviewer R2 (green square) had a reduced sensitivity and FDR with CDSS. The REMI automated seizure detection algorithm (black diamond) had a higher sensitivity than any one reviewer or consensus, with a higher false detection rate. The error bars are 95% confidence intervals using a bias-corrected and accelerated method (BCa, N = 1,000).
Article Snippet: Epitel's automated seizure detection algorithm was able to detect 21 of 24 known focal-onset seizures (out-of-sample), providing a mean sensitivity of 90% across all subjects (100% for 8 out of 10 subjects, —Algorithm, —filled black diamond; ).
Techniques: Software