EasyBCI Agent: Towards Universal Neural Data Preprocessing for Brain-Computer Interfaces
Abstract
Brain-computer interfaces translate neural activity into device commands, yet their performance hinges on preprocessing that remains manual, expert-dependent and poorly reproducible. Large language model agents can automate scientific coding, but existing systems lack the modality coverage, raw-data isolation, experience accumulation and domain oversight that neural preprocessing requires. We introduce EasyBCI, a two-phase LLM agent that plans and executes preprocessing pipelines for six signal types. A Plan Agent profiles each recording into a text-only Data Fingerprint that never exposes raw data to the model and selects a literature-grounded operator sequence. An Execution Agent generates, runs and self-corrects code until quality criteria are met, while a quality-gated experience system retains validated strategies as reusable skills. A domain expert intervenes at two decision gates, retaining human judgement where undetected error can invalidate downstream analyses. Evaluation on EEG with a fixed linear classifier shows that all five EasyBCI backbones preserve more task-relevant separability than the manual pipeline. Under same-backbone comparison, EasyBCI outperforms general-purpose coding agents on both label schemes for four of five configurations. EasyBCI extends to five additional modalities spanning nearly three orders of magnitude in sampling rate, producing complete reproducible pipelines with recorded decision provenance. These results indicate that domain-specific orchestration can bring auditable preprocessing within reach of laboratories lacking dedicated expertise, illustrating design principles applicable to AI agents in other scientific domains.
Related
- Generalizable neural decoding for clinically scalable brain-computer interfaces.
- Assistive algorithms influence neural representations in motor brain-computer interfaces
- Slow-Fast Brain-Computer Interfaces: Preventing Neuroadaptive Overfitting in AI-Mediated Neural Interfaces
- Learning Dynamic Neural Evidence Representations for Time-Adaptive Brain-Computer Interfaces
- Engineering Neural Interface Materials for Brain–Computer Interfaces: Toward Reliable Neural Information and Functional Recovery
BibTeX
@misc{zhu2026easybci,
title = {{EasyBCI Agent: Towards Universal Neural Data Preprocessing for Brain-Computer Interfaces}},
author = {Yu Zhu and Runkai Zhao and Zhimin Zhou and Jinyu Cai and Zhouheng Yao and Chutian Zhang and Peiyuan Li and Xiaoxing Zhang and Qihao Zheng and Jiamin Wu and Mianxin Liu and Chi Zhang and Bo Zhang and Bin Min and Lei Bai and Chengyu Li and Jingpeng Wu and Chunfeng Song},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2607.29007},
eprint = {2607.29007},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.29007},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - EasyBCI Agent: Towards Universal Neural Data Preprocessing for Brain-Computer Interfaces
AU - Yu Zhu
AU - Runkai Zhao
AU - Zhimin Zhou
AU - Jinyu Cai
AU - Zhouheng Yao
AU - Chutian Zhang
AU - Peiyuan Li
AU - Xiaoxing Zhang
AU - Qihao Zheng
AU - Jiamin Wu
AU - Mianxin Liu
AU - Chi Zhang
AU - Bo Zhang
AU - Bin Min
AU - Lei Bai
AU - Chengyu Li
AU - Jingpeng Wu
AU - Chunfeng Song
PY - 2026
DA - 2026/09/01
PB - arXiv
DO - 10.48550/arxiv.2607.29007
UR - https://arxiv.org/abs/2607.29007
AB - Brain-computer interfaces translate neural activity into device commands, yet their performance hinges on preprocessing that remains manual, expert-dependent and poorly reproducible. Large language model agents can automate scientific coding, but existing systems lack the modality coverage, raw-data isolation, experience accumulation and domain oversight that neural preprocessing requires. We introduce EasyBCI, a two-phase LLM agent that plans and executes preprocessing pipelines for six signal types. A Plan Agent profiles each recording into a text-only Data Fingerprint that never exposes raw data to the model and selects a literature-grounded operator sequence. An Execution Agent generates, runs and self-corrects code until quality criteria are met, while a quality-gated experience system retains validated strategies as reusable skills. A domain expert intervenes at two decision gates, retaining human judgement where undetected error can invalidate downstream analyses. Evaluation on EEG with a fixed linear classifier shows that all five EasyBCI backbones preserve more task-relevant separability than the manual pipeline. Under same-backbone comparison, EasyBCI outperforms general-purpose coding agents on both label schemes for four of five configurations. EasyBCI extends to five additional modalities spanning nearly three orders of magnitude in sampling rate, producing complete reproducible pipelines with recorded decision provenance. These results indicate that domain-specific orchestration can bring auditable preprocessing within reach of laboratories lacking dedicated expertise, illustrating design principles applicable to AI agents in other scientific domains.
ER -