MindType

The Project

Neurodegenerative disabilities such as Amyotrophic Lateral Sclerosis (ALS) cause the degeneration of motor neurons, leading to severe muscle weakness. As a result, many ALS patients become socially isolated and unable to communicate independently using conventional keyboards. This project developed MindType, an automated brain-computer interface (BCI) typing system that enables users to input text directly from electroencephalogram (EEG) signals without physical movement. The system employs a Steady-State Visual Evoked Potential (SSVEP) paradigm in which a virtual keyboard presents characters flickering at distinct frequencies. Non-invasive EEG signals were collected while participants focused on target keys, preprocessed to remove noise, and then analyzed using a BiLSTM neural network trained to decode users’ intended characters from high-dimensional neural patterns. Experimental evaluation involved repeated typing trials. During testing, the system achieved a mean accuracy of 89.30% ± 5.55%, and an information transfer rate (ITR) of 32 bits/min. These results validate the effectiveness of MindType as a practical EEG-based typing solution. The high accuracy, stability, and throughput demonstrate that BiLSTM-based decoding of SSVEP signals can provide reliable, real-time communication for individuals with ALS or severe motor disabilities.

View this project on YouTube