The study introduces a meta-learning and pretraining approach to improve neural stimulation response modeling. By extending temporal basis function models with a MAML-based architecture, the authors demonstrate a significant reduction in catastrophic forecast failures and narrower prediction intervals across 40 optogenetic stimulation sessions in non-human primates. The method also cuts calibration requirements by 50–90%, making closed‑loop stimulation more feasible within clinical time constraints.
By Matthew J Bryan, Daniel C Muir, Felix Schwock, Azadeh Yazdan-Shahmorad, Rajesh P N Rao
arXiv:2609.36087v1 Announce Type: cross
Abstract: Clinicians and neuroscientists have long analyzed intracranial electroencephalography (iEEG) through directly measurable physiological characteristic...
By Yipeng Zhang, Chenda Duan, Yuanyi Ding, Tianyi Wang, Atsuro Daida, Masaki Izumi, Yuta Tanoue, Naoto Kuroda, Shaun A. Hussain, Nishant Sinha, Eishi Asano. Hiroki Nariai, Vwani Roychowdhury
arXiv:2608.24727v1 Announce Type: cross
Abstract: EEG foundation models pretrained via self-supervised learning promise transferable representations, but their generalization remains limited, especia...
By Meghal Dani, Stefanie Liebe
arXiv:2606. 00074v1 Announce Type: cross Abstract: Reliable seizure prediction is a prerequisite for closed-loop neurostimulation therapy, yet existing methods rarely account for the variability in EEG signal quality encountered in real-world deployment, and the overwhelming majority adopt non-strict evaluation protocols that overestimate generalisation performance.
By Mufeng Chen, Qi Wu, Bingchao Huang, Xiwen Lai, Zekai Chen, Xinge Ouyang, Quansheng Ren
arXiv:2609.27729v1 Announce Type: cross
Abstract: In neuropsychiatry, the primary goal is often not only to decode brain activity but to change it, for example to lessen a negative affective bias or...
By Marco Rothermel, Madleen Stenger, Soroush Daftarian, Svenja Jule Francke, Bita Shariatpanahi, Jos\'e C. Garc\'ia Alanis, Mohammad-Ali Nikouei Mahani, Stefan G. Hofmann, Tim Hahn, Hamidreza Jamalabadi
The paper demonstrates that meta‑learning and pretraining can improve neural stimulation response models, reducing catastrophic forecast failures and narrowing prediction intervals. Using temporal basis function models with a MAML‑based architecture, the authors evaluated 40 optogenetic stimulation sessions in non‑human primates and found that a 1,000‑sample calibration set reduced poor‑performance sessions from 16 to 1. Calibration needs were cut by 50–90%, making closed‑loop stimulation feasible within clinical time limits.