Search papers, labs, and topics across Lattice.
This paper introduces NeuraDock Agent, an open-source architecture designed to enhance the usability of low-channel EEG data by integrating a deterministic local EEG engine with a hardware-aware language layer. By providing the LLM with a compact summary and a versioned context pack that details hardware specifications and implementation boundaries, the system ensures that interpretations of EEG data remain grounded in the actual capabilities and limitations of the hardware. Evaluation results demonstrate the system's reliability through consistent output across multiple repetitions and successful handling of data boundaries under various experimental conditions, highlighting the importance of context-aware grounding in EEG analysis.
Grounding EEG interpretations in hardware capabilities can drastically reduce unsupported conclusions and improve the reliability of scientific software.
Large language models (LLMs) can make scientific software easier to use. However, a general model does not automatically know which measurements a particular sensor can support, which algorithms are implemented in the current software, or which conclusions are justified by a computed result. These distinctions are especially important for low-channel electroencephalography (EEG), where sparse spatial coverage and variable signal quality make plausible but unsupported interpretations easy to produce. We present NeuraDock Agent, an open-source architecture that separates a deterministic local EEG engine from a hardware-aware language layer. The numerical engine parses recordings, performs quality control, executes reviewed spectral workflows, and writes machine-readable artifacts. The LLM receives only a compact, allowlisted summary and a versioned context pack. The context describes the seven-channel hardware, reviewed workflows, result fields, implementation boundaries, scientific limits, and reference cases. Raw EEG and dense per-sample arrays remain local We evaluate the system at three levels. First, 12 recordings produced identical structured results over ten numerical repetitions, and a complete Rest/Task run produced identical result, report, and figure hashes over three repetitions. Second, request-capture and failure-injection experiments confirmed the tested data boundary and preservation of local artifacts under HTTP, malformed-output, and connection failures. Third, a boundary-awareness benchmark tested 36 ordinary and adversarial questions under four context ablations and two LLMs, yielding 288 outputs.These results support hardware- and implementation-aware grounding as a practical mechanism for calibrating what an EEG agent accepts, qualifies, or refuses; they do not establish clinical validity or a validated absolute cognitive-load index.
