Abstract
<title>Abstract</title> <p>Clinical electroencephalography (EEG) reporting remains largely manual and time-consuming, and current EEG software ecosystems are not designed to pro duce the structured EEG–text supervision needed for training modern language models. Most toolboxes focus on visualization, preprocessing, or event marking, but provide limited support for clinician-centred workflows that simultaneously generate high-quality datasets for AI. We introduce EEG-to-Report, a browser based annotation and feature–text framework that links routine EEG review with the construction of AI-ready datasets. The framework integrates multi-format EEG ingestion with automatic channel standardisation and preprocessing, an interactive multi-channel viewer for drag-based time-range and channel selection, and a multimodal annotation layer that combines typed text with voice notes transcribed via speech-to-text. For each annotated segment, a feature extrac tion engine computes a standardised set of spectral, temporal, entropy, Hjorth, connectivity, and spike-related descriptors, which are stored together with the corresponding clinical descriptions in a portable JSON schema capturing seg ment timing, channel context, features and notes. This representation yields aligned feature–text pairs designed to supervise multimodal EEG–language mod els. As a working report generator, the framework includes an in-application auto-report module that combines an ensemble of convolutional networks with a large language model to draft clinical narratives; training a feature-to-text model on the exported corpus is the natural next step, which we outline as future work. Using pilot annotations, we describe how EEG-to-Report is designed to streamline annotation workflows and produce editable draft reports for review by neurologists, providing a reusable foundation for future EEG–text corpora and automated EEG reporting systems.</p>