2021/10/26 by Jens Müller, Hongliu Yang, Müller, Jens +11
Medicine · Neuroscience · #EEG and Brain-Computer Interfaces #Epilepsy research and treatment #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Signal Processing (eess.SP) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2110.13550
openalex publication_date 2021/10/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Seizure forecasting using machine learning is possible, but the performance\nis far from ideal, as indicated by many false predictions and low specificity.\nHere, we examine false and missing alarms of two algorithms on long-term\ndatasets to show that the limitations are less related to classifiers or\nfeatures, but rather to intrinsic changes in the data. We evaluated two\nalgorithms on three datasets by computing the correlation of false predictions\nand estimating the information transfer between both classification methods.\nFor 9 out of 12 individuals both methods showed a performance better than\nchance. For all individuals we observed a positive correlation in predictions.\nFor individuals with strong correlation in false predictions we were able to\nboost the performance of one method by excluding test samples based on the\nresults of the second method. Substantially different algorithms exhibit a\nhighly consistent performance and a strong coherency in false and missing\nalarms. Hence, changing the underlying hypothesis of a preictal state of fixed\ntime length prior to each seizure to a proictal state is more helpful than\nfurther optimizing classifiers. The outcome is significant for the evaluation\nof seizure prediction algorithms on continuous data.\n