A systematic review of the efficiency of P300: based digital spellers

Compartilhe:

Abstract:

The integration of computing and medicine has significantly advanced assistive communication technologies for individuals with neurodegenerative diseases. This systematic review comprehensively examines the performance and bit rate of P300-based digital spellers, with a particular focus on the application of machine learning techniques. Evaluating these spellers’ bit rates is crucial for their practical deployment. The review analyzes various visual paradigms and classification algorithms employed to enhance the spellers’ performance, emphasizing the trade-off between accuracy and bit rate. Key findings indicate that while the row/column paradigm is commonly used due to its high accuracy, reaching up to 97.6%, alternatives such as three-dimensional matrices and singlecharacter paradigms demonstrate greater potential, achieving bit rates up to 66 bits per minute. The choice of classification algorithm, particularly variants of Linear Discriminant Analysis (LDA), plays a pivotal role in system performance. Notably, Bayesian LDA (BLDA) and Convolutional Neural Networks (CNN) have shown accuracies of approximately 91% and 95.82%, respectively. Additionally, individual differences, such as age and neurological conditions, significantly affect the system’s efficiency.

Referência:

NOVAIS, Victor Hugo Gomes de; LEAL, Adriano Galindo. A systematic review of the efficiency of P300: based digital spellers. In: INTERNATIONAL CONFERENCE ON ADVANCES IN ARTIFICIAL INTELLIGENCE, 8., 2024, London. Proceendings… 7p.

Documento com acesso restrito. Logar na BiblioInfo, Biblioteca GITEB/IPT para acessar o artigo em PDF:

https://escriba.ipt.br/pdf_restrito/179375.pdf

INSCREVA-se em nossa newsletter

Receba nossas novidades em seu e-mail.

SUBSCRIBE to our newsletter

Receive our news in your email.

Pular para o conteúdo