MindChat: Enhancing BCI Spelling with Large Language Models in Realistic Scenarios

TL;DR

MindChat leverages large language models to improve BCI spelling efficiency by reducing keystrokes and spelling time, making communication more practical for users with motor disabilities.

Contribution

This paper introduces MindChat, a novel LLM-assisted BCI speller that uses prompt engineering to provide context-aware predictions, significantly enhancing spelling efficiency.

Findings

MindChat reduces keystrokes by over 62%

Spelling time decreases by more than 32%

Effective in four dialogue scenarios

Abstract

Brain-computer interface (BCI) spellers can render a new communication channel independent of peripheral nervous system, which are especially valuable for patients with severe motor disabilities. However, current BCI spellers often require users to type intended utterances letter-by-letter while spelling errors grow proportionally due to inaccurate electroencephalogram (EEG) decoding, largely impeding the efficiency and usability of BCIs in real-world communication. In this paper, we present MindChat, a large language model (LLM)-assisted BCI speller to enhance BCI spelling efficiency by reducing users' manual keystrokes. Building upon prompt engineering, we prompt LLMs (GPT-4o) to continuously suggest context-aware word and sentence completions/predictions during spelling. Online copy-spelling experiments encompassing four dialogue scenarios demonstrate that MindChat saves more than 62\% keystrokes and over 32\% spelling time compared with traditional BCI spellers. We envision high-speed BCI spellers enhanced by LLMs will potentially lead to truly practical applications.