Relação entre desempenho e conforto em interfaces cérebro-máquina: análise paramétrica em potenciais evocados visuais

Abstract

Brain-Computer Interfaces (BCIs) have a wide range of applications, such as controlling robotic hands, rehabilitating patients with limb paralysis caused by strokes, attention training for individuals with attention deficit, and controlling alternative communication interfaces. Each application involves recognizing different sets of brain activities, such as Motor Imagery (MI), attention, and Steady-State Visual Evoked Potentials (SSVEP). In this study, the SSVEP paradigm will be analyzed. These signals arise after visual stimuli at specific frequencies, generating synchronized brain waves. Classifying the generated waves into their respective frequencies enables the use of these systems for controlling continuous or discrete interfaces solely through the subject’s visual focus. These systems have an advantage, as they typically do not require any model training for operation, unlike models for Motor Imagery or attention. Although they achieve high accuracy and Information Transfer Rates (ITR), these applications induce significant visual fatigue in users. This fatigue is a recurring topic in SSVEP application studies; however, few analyses consider frequency selection concerning user comfort. For this reason, this study examines different time and frequency windows concerning the model’s performance and the subjects’ average comfort. Furthermore, analyzing different target quantities allows the evaluation of the model’s usability in various applications, such as typing systems, mechanical arm control, and interaction in digital games. Additionally, an analysis focused on ITR identified a set of parameters that stand out in both comfort and performance, providing options that can prioritize either system efficiency or user experience.