Learning to generalize seizure forecasts
© 2022 The Authors. Epilepsia published by Wiley Periodicals LLC on behalf of International League Against Epilepsy..
OBJECTIVE: Epilepsy is characterized by spontaneous seizures that recur at unexpected times. Nonetheless, using years-long electroencephalographic (EEG) recordings, we previously found that patient-reported seizures consistently occur when interictal epileptiform activity (IEA) cyclically builds up over days. This multidien (multiday) interictal-ictal relationship, which is shared across patients, may bear phasic information for forecasting seizures, even if individual patterns of seizure timing are unknown. To test this rigorously in a large retrospective dataset, we pretrained algorithms on data recorded from a group of patients, and forecasted seizures in other, previously unseen patients.
METHODS: We used retrospective long-term data from participants (N = 159) in the RNS System clinical trials, including intracranial EEG recordings (icEEG), and from two participants in the UNEEG Medical clinical trial of a subscalp EEG system (sqEEG). Based on IEA detections, we extracted instantaneous multidien phases and trained generalized linear models (GLMs) and recurrent neural networks (RNNs) to forecast the probability of seizure occurrence at a 24-h horizon.
RESULTS: With GLMs and RNNs, seizures could be forecasted above chance in 79% and 81% of previously unseen subjects with a median discrimination of area under the curve (AUC) = .70 and .69 and median Brier skill score (BSS) = .07 and .08. In direct comparison, individualized models had similar median performance (AUC = .67, BSS = .08), but for fewer subjects (60%). Moreover, calibration of pretrained models could be maintained to accommodate different seizure rates across subjects.
SIGNIFICANCE: Our findings suggest that seizure forecasting based on multidien cycles of IEA can generalize across patients, and may drastically reduce the amount of data needed to issue forecasts for individuals who recently started collecting chronic EEG data. In addition, we show that this generalization is independent of the method used to record seizures (patient-reported vs. electrographic) or IEA (icEEG vs. sqEEG).
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2023 |
|---|---|
| Erschienen: |
2023 |
| Enthalten in: |
Zur Gesamtaufnahme - volume:64 Suppl 4 |
|---|---|
| Enthalten in: |
Epilepsia - 64 Suppl 4(2023) vom: 05. Dez., Seite S99-S113 |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Leguia, Marc G [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
Intracranial EEG |
|---|
| Anmerkungen: |
Date Completed 28.12.2023 Date Revised 28.12.2023 published: Print-Electronic Citation Status MEDLINE |
|---|
| doi: |
10.1111/epi.17406 |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM345903382 |
|---|
| LEADER | 01000caa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM345903382 | ||
| 003 | DE-627 | ||
| 005 | 20231229124058.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 231226s2023 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1111/epi.17406 |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1241.xml |
| 035 | |a (DE-627)NLM345903382 | ||
| 035 | |a (NLM)36073237 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Leguia, Marc G |e verfasserin |4 aut | |
| 245 | 1 | 0 | |a Learning to generalize seizure forecasts |
| 264 | 1 | |c 2023 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ƒaComputermedien |b c |2 rdamedia | ||
| 338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Date Completed 28.12.2023 | ||
| 500 | |a Date Revised 28.12.2023 | ||
| 500 | |a published: Print-Electronic | ||
| 500 | |a Citation Status MEDLINE | ||
| 520 | |a © 2022 The Authors. Epilepsia published by Wiley Periodicals LLC on behalf of International League Against Epilepsy. | ||
| 520 | |a OBJECTIVE: Epilepsy is characterized by spontaneous seizures that recur at unexpected times. Nonetheless, using years-long electroencephalographic (EEG) recordings, we previously found that patient-reported seizures consistently occur when interictal epileptiform activity (IEA) cyclically builds up over days. This multidien (multiday) interictal-ictal relationship, which is shared across patients, may bear phasic information for forecasting seizures, even if individual patterns of seizure timing are unknown. To test this rigorously in a large retrospective dataset, we pretrained algorithms on data recorded from a group of patients, and forecasted seizures in other, previously unseen patients | ||
| 520 | |a METHODS: We used retrospective long-term data from participants (N = 159) in the RNS System clinical trials, including intracranial EEG recordings (icEEG), and from two participants in the UNEEG Medical clinical trial of a subscalp EEG system (sqEEG). Based on IEA detections, we extracted instantaneous multidien phases and trained generalized linear models (GLMs) and recurrent neural networks (RNNs) to forecast the probability of seizure occurrence at a 24-h horizon | ||
| 520 | |a RESULTS: With GLMs and RNNs, seizures could be forecasted above chance in 79% and 81% of previously unseen subjects with a median discrimination of area under the curve (AUC) = .70 and .69 and median Brier skill score (BSS) = .07 and .08. In direct comparison, individualized models had similar median performance (AUC = .67, BSS = .08), but for fewer subjects (60%). Moreover, calibration of pretrained models could be maintained to accommodate different seizure rates across subjects | ||
| 520 | |a SIGNIFICANCE: Our findings suggest that seizure forecasting based on multidien cycles of IEA can generalize across patients, and may drastically reduce the amount of data needed to issue forecasts for individuals who recently started collecting chronic EEG data. In addition, we show that this generalization is independent of the method used to record seizures (patient-reported vs. electrographic) or IEA (icEEG vs. sqEEG) | ||
| 650 | 4 | |a Journal Article | |
| 650 | 4 | |a intracranial EEG | |
| 650 | 4 | |a multidien | |
| 650 | 4 | |a seizure forecasting | |
| 650 | 4 | |a subscalp EEG | |
| 650 | 4 | |a transfer learning | |
| 700 | 1 | |a Rao, Vikram R |e verfasserin |4 aut | |
| 700 | 1 | |a Tcheng, Thomas K |e verfasserin |4 aut | |
| 700 | 1 | |a Duun-Henriksen, Jonas |e verfasserin |4 aut | |
| 700 | 1 | |a Kjaer, Troels W |e verfasserin |4 aut | |
| 700 | 1 | |a Proix, Timothée |e verfasserin |4 aut | |
| 700 | 1 | |a Baud, Maxime O |e verfasserin |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Epilepsia |d 1954 |g 64 Suppl 4(2023) vom: 05. Dez., Seite S99-S113 |w (DE-627)NLM000016195 |x 1528-1167 |7 nnns |
| 773 | 1 | 8 | |g volume:64 Suppl 4 |g year:2023 |g day:05 |g month:12 |g pages:S99-S113 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1111/epi.17406 |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |d 64 Suppl 4 |j 2023 |b 05 |c 12 |h S99-S113 | ||