Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386875692> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4386875692 abstract "Objective: Forecasting epileptic seizures can reduce uncertainty for patients and allow preventative actions. While many models can predict the occurrence of seizures from features of the EEG, few models incorporate changes in features over time. Long Short-Term Memory (LSTM) neural networks are a machine learning architecture that can display temporal dynamics due to the recurrent connections. In this paper, we used LSTMs to monitor changes in EEG features over time to improve the accuracy of seizure forecasts and to alter the time window of the forecast. Methods: Long-term intracranial EEG recordings from eight patients from the NeuroVista dataset were used. A Fourier transform of 1-minute segments of EEG was fed into a Convolutional Neural Network (CNN). The outputs from the CNN were input to three different LSTM models at different time intervals: 1 minute, 1 hour and 1 day. The LSTM model outputs were used to predict seizure onset within a time window. The prediction and start of the time window were separated by the same length of time as the window. Window sizes tested included 2, 4, 10, 20 and 40 minutes. Results and Conclusion: Our model forecast seizure onsets well above a random predictor. Compared to other models using the same dataset, our model performed better for some patients and worse for others. Monitoring the change in EEG features over time allowed our model to produce good results over a range of different window sizes, which is an improvement on previous models and raises the possibility of altering the forecast to meet individual patient needs. Furthermore, a window size of 40 minutes provides a potential intervention time of 40 minutes, which is the first time an intervention time of more than 5 minutes have been forecast using long-term EEG recordings." @default.
- W4386875692 created "2023-09-20" @default.
- W4386875692 creator A5002748431 @default.
- W4386875692 creator A5014100961 @default.
- W4386875692 creator A5017711423 @default.
- W4386875692 creator A5055017686 @default.
- W4386875692 creator A5065275567 @default.
- W4386875692 creator A5066512588 @default.
- W4386875692 creator A5073857507 @default.
- W4386875692 date "2023-09-18" @default.
- W4386875692 modified "2023-09-27" @default.
- W4386875692 title "Epileptic seizure forecasting with long short-term memory (LSTM) neural networks" @default.
- W4386875692 doi "https://doi.org/10.48550/arxiv.2309.09471" @default.
- W4386875692 hasPublicationYear "2023" @default.
- W4386875692 type Work @default.
- W4386875692 citedByCount "0" @default.
- W4386875692 crossrefType "posted-content" @default.
- W4386875692 hasAuthorship W4386875692A5002748431 @default.
- W4386875692 hasAuthorship W4386875692A5014100961 @default.
- W4386875692 hasAuthorship W4386875692A5017711423 @default.
- W4386875692 hasAuthorship W4386875692A5055017686 @default.
- W4386875692 hasAuthorship W4386875692A5065275567 @default.
- W4386875692 hasAuthorship W4386875692A5066512588 @default.
- W4386875692 hasAuthorship W4386875692A5073857507 @default.
- W4386875692 hasBestOaLocation W43868756921 @default.
- W4386875692 hasConcept C111919701 @default.
- W4386875692 hasConcept C119857082 @default.
- W4386875692 hasConcept C121332964 @default.
- W4386875692 hasConcept C147168706 @default.
- W4386875692 hasConcept C153180895 @default.
- W4386875692 hasConcept C154945302 @default.
- W4386875692 hasConcept C15744967 @default.
- W4386875692 hasConcept C169760540 @default.
- W4386875692 hasConcept C2778751112 @default.
- W4386875692 hasConcept C2779334592 @default.
- W4386875692 hasConcept C41008148 @default.
- W4386875692 hasConcept C50644808 @default.
- W4386875692 hasConcept C522805319 @default.
- W4386875692 hasConcept C61797465 @default.
- W4386875692 hasConcept C62520636 @default.
- W4386875692 hasConcept C81363708 @default.
- W4386875692 hasConceptScore W4386875692C111919701 @default.
- W4386875692 hasConceptScore W4386875692C119857082 @default.
- W4386875692 hasConceptScore W4386875692C121332964 @default.
- W4386875692 hasConceptScore W4386875692C147168706 @default.
- W4386875692 hasConceptScore W4386875692C153180895 @default.
- W4386875692 hasConceptScore W4386875692C154945302 @default.
- W4386875692 hasConceptScore W4386875692C15744967 @default.
- W4386875692 hasConceptScore W4386875692C169760540 @default.
- W4386875692 hasConceptScore W4386875692C2778751112 @default.
- W4386875692 hasConceptScore W4386875692C2779334592 @default.
- W4386875692 hasConceptScore W4386875692C41008148 @default.
- W4386875692 hasConceptScore W4386875692C50644808 @default.
- W4386875692 hasConceptScore W4386875692C522805319 @default.
- W4386875692 hasConceptScore W4386875692C61797465 @default.
- W4386875692 hasConceptScore W4386875692C62520636 @default.
- W4386875692 hasConceptScore W4386875692C81363708 @default.
- W4386875692 hasLocation W43868756921 @default.
- W4386875692 hasOpenAccess W4386875692 @default.
- W4386875692 hasPrimaryLocation W43868756921 @default.
- W4386875692 hasRelatedWork W2097120763 @default.
- W4386875692 hasRelatedWork W2544144554 @default.
- W4386875692 hasRelatedWork W2767651786 @default.
- W4386875692 hasRelatedWork W2912288872 @default.
- W4386875692 hasRelatedWork W3021430260 @default.
- W4386875692 hasRelatedWork W3027997911 @default.
- W4386875692 hasRelatedWork W3198971141 @default.
- W4386875692 hasRelatedWork W4287776258 @default.
- W4386875692 hasRelatedWork W4316077036 @default.
- W4386875692 hasRelatedWork W2296457990 @default.
- W4386875692 isParatext "false" @default.
- W4386875692 isRetracted "false" @default.
- W4386875692 workType "article" @default.