Cutting-Edge Technology Gives Hope to Non-speaking Patients

A revolutionary technology has been developed with the potential to decipher the thoughts of paralysed individuals using artificial intelligence. This innovation could revolutionise communication for those unable to speak due to various conditions, such as stroke or paralysis. Furthermore, the AI has the capability to enable seamless communication between humans and machines, such as the operation of robots or bionic limbs.

The technology operates through a portable device that detects brainwaves via a sensory cap, without the need for invasive implants or surgical procedures. In a trial conducted by researchers at the University of Technology Sydney’s GrapheneX-UTS Human-Centric AI Centre, participants wore a cap that recorded electrical brain activity through their scalp using an electroencephalogram (EEG) while silently reading passages of text. The device, known as DeWave, then segmented each person’s EEG wave to capture specific brain patterns and characteristics. By studying large quantities of EEG data, the AI was able to translate EEG signals into words and sentences.

This innovation distinguishes itself from previous brain signal to language technologies, such as Neuralink or MRI machines. Unlike Neuralink, which necessitates invasive surgery to implant electrodes into the brain, and the MRI machine, which is large, expensive, and impractical for daily use, the new technology offers a non-invasive and potentially safer alternative. Additionally, the new system is capable of transforming brain signals into individual word segments without the need for additional aids, such as eye-tracking, which limits practical application.

According to Distinguished Professor Chin-Teng Lin, the director of the study, this research represents a significant advancement in the field, integrating discrete encoding techniques in the brain-to-text translation process and introducing new frontiers in neuroscience and AI by integrating with large language models. The involvement of 29 participants in the study makes the technology more robust and adaptable, as EEG waves can vary significantly between individuals.

While the current accuracy score for translation is approximately 40 per cent, researchers aim to enhance this over time to a level comparable to traditional language translation or speech recognition programmes, which typically score closer to 90 per cent. Lead author Dr Yigun Duan highlighted that the model is more accurate at matching verbs than nouns, with a tendency towards synonymous pairs for nouns, such as ‘the man’ instead of ‘the author’, which he attributes to the similar brain wave patterns produced by semantically related words. Regardless of these challenges, the model produces meaningful results by aligning keywords and forming similar sentence structures.

The study has been recognised as the spotlight paper at the NeurIPS conference, a prestigious annual meeting showcasing world-leading research on AI and machine learning. This breakthrough follows on from previous brain-computer interface technology developed by the University of Technology Sydney, which utilised brainwaves to command a quadruped robot in collaboration with the Australian Defence Force.

In conclusion, this cutting-edge technology provides hope for individuals unable to communicate verbally due to illness or injury, and represents a significant step forward in the fields of neuroscience and AI.