Creating Mathematical and Machine Learning Models to Understand the Brain And Epilepsy

This PhD thesis tackles the complex challenge of understanding the brain and epilepsy, a task constrained by the limitations of current neurophysiological and brain imaging techniques. It introduces innovative machine learning and mathematical models applied to iEEG data, aiming to enhance interpretability and predictability in epilepsy research. The LSTM filter as one of the key developments, integrated with the NMM, surpasses traditional methods like the Kalman Filter in accuracy and efficiency. Combined with the seizure duration prediction, and machine learning methods, it can also show the spatial-temporal change and how a seizure can evolve. Furthermore, the thesis advances seizure prediction using machine learning models that analyse critical slowing down and other features.