| Popis: |
OBJECTIVE Epilepsy is characterized by spontaneous seizures that recur at unexpected times. Yet, using years-long EEG recordings, we previously found that patient-reported seizures consistently occur when interictal epileptiform activity (IEA) cyclically builds up over days. This multidien (multi-day) 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 pre-trained 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 clinical trial of a sub-scalp 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-hour 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 AUC=0.70 and 0.69 and median Brier skill score of BSS=0.07 and 0.08. In direct comparison, individualized models had similar median performance (AUC=0.67, BSS=0.08), but for fewer subjects (60%). Moreover, calibration of pre-trained 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). |
Plný text