What does this research mean for the field?

ChatBCI reduces keystrokes and accelerates sentence composition in P300 speller BCIs by leveraging large language models for word prediction. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

The aim is to improve the efficiency and user-friendliness of P300 speller BCIs for sentence composition.

February 19, 2026Open Access

ChatBCI, a P300 speller BCI with context-driven word prediction leveraging large language models, from concept to evaluation

Key Points

The aim is to improve the efficiency and user-friendliness of P300 speller BCIs for sentence composition.
Developed ChatBCI utilizing large language models for word prediction.
Implemented a modified GUI with GPT-3.5 word suggestions.
Applied stepwise linear discriminant analysis for P300 classification.
Conducted online spelling tasks with seven subjects for testing.
Compared performance between traditional letter-by-letter BCIs and ChatBCI.
ChatBCI significantly reduces time and keystrokes for sentence composition.
In copy-spelling tasks, ChatBCI outperforms traditional spellers in keystroke savings and information transfer rates.
Improvised sessions with ChatBCI yielded substantial keystroke savings across subjects.

Abstract

P300 speller brain computer interfaces (BCIs) allow users to compose sentences by selecting target keys on a graphical user interface (GUI) through the detection of P300 component in their electroencephalogram (EEG) signals following visual stimuli. Most existing P300 speller BCIs require users to spell all or the first few initial letters of the intended word, letter by letter. Consequently, a large number of keystrokes could be required to write an intended sentence, thereby, increasing user's time and cognitive load. There is a need for more efficient and user-friendly methods for faster, and practical sentence composition. In this work, we introduce ChatBCI, a P300 speller BCI that leverages the zero-shot learning capabilities of large language models (LLMs) to suggest words from user-spelled initial letters or predict the subsequent word(s), reducing keystrokes and accelerating sentence composition. ChatBCI retrieves word suggestions through remote queries to the GPT-3.5 API. A modified GUI, displaying GPT-3.5 word suggestions as extra keys is designed. Stepwise linear discriminant analysis (SWLDA) is used for the P300 classification. Seven subjects completed two online spelling tasks: 1) copy-spelling a self-composed sentence using ChatBCI, and 2) improvising a sentence using ChatBCI's word suggestions. Results demonstrate that for the copy-spelling task, on average, ChatBCI outperforms letter-by-letter BCI spellers, reducing time and keystrokes by Formula: see text and Formula: see text, respectively, and increasing information transfer rate by Formula: see text. For the improvised sessions, ChatBCI achieves Formula: see text keystroke savings across subjects. Overall, ChatBCI, by employing remote LLM queries outperforms traditional spellers without requiring local model training or storage. ChatBCI's (multi-)word prediction capability paves the way for developing next-generation speller BCIs that are efficient and effective for real-time communication, specially for users with communication and motor disabilities.

Bookmark

View Full Paper