EEG-to-Report Framework Builds Training Data for Clinical EEG Language Models

Summary

EEG-to-Report is a browser-based framework for creating structured, AI-ready clinical EEG datasets. It combines EEG ingestion, channel standardization, interactive review, typed and transcribed annotations, and standardized signal features in a portable JSON schema. These aligned feature-text pairs are designed to train multimodal EEG-language models. The framework also combines convolutional networks with a large language model to draft editable clinical narratives for neurologist review. Pilot annotations indicate that it can streamline dataset creation and support future automated EEG reporting systems, while larger clinical validation is still needed.