Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
November 1, 2021

Random Forest classifiers achieved up to 94.3% ± 5.3% accuracy in classifying absence from tonic-clonic seizures using reduced EEG timeseries, outperforming Support Vector Machines.

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Wearable systems have few electrodes and reduced spatial resolution compared to full-scalp EEG, creating a need to evaluate seizure classification performance using reduced timeseries.

Does Random Forest classification improve seizure-type classification accuracy compared to SVM using reduced EEG timeseries?

Population

Patients with absence and tonic-clonic seizures

Comparison

Random Forest classifiers vs Support Vector Machines and Decision Trees across reduced EEG subsets

Key result

Random Forest classifiers achieved up to 94.3% ± 5.3% accuracy in classifying absence from tonic-clonic seizures using reduced EEG timeseries, outperforming Support Vector Machines.

Authors

SNSébastien NazeJTJianbin TangJKJames Kozloski

Discussion

Loading...

Member takes

Overview

May support wearable EEG systems for seizure classification; hypothesis-generating and should not change practice without validation.

Structured PICO

Does Random Forest classification improve seizure-type classification accuracy compared to SVM using reduced EEG timeseries?

P
Population
EEG recordings from patients with absence and tonic-clonic seizures
E
Exposure
Random Forest (RF) classifiers using a subset of EEG recordings (single trace or dimensionality reduction)
C
Comparator
Support Vector Machines (SVM) and Decision Trees
O
Outcome
Seizure-type classification accuracy (absence vs tonic-clonic)surrogate

Random Forest classifiers provide efficient and stable seizure classification using reduced EEG channels, suitable for embedded wearable systems.

Cite This Study

Naze et al. (2021) studied Seizures. Random Forest (RF) classifiers using reduced EEG timeseries vs. Support Vector Machines (SVM) was evaluated on Accuracy in classifying absence from tonic-clonic seizures. Random Forest classifiers achieved up to 94.3% ± 5.3% accuracy in classifying absence from tonic-clonic seizures using reduced EEG timeseries, outperforming Support Vector Machines.

synapsesocial.com/papers/6a22b855cce2ba38c0cd6872https://doi.org/10.1109/embc46164.2021.9630398
View Full Paper
Ask AI
Bookmark
Share