Aplicación de técnicas de aprendizaje de máquina para facilitar al comunicación en pacientes con neuropatía periférica utilizando señáles EEG y visión computacional : Desarrollo de un modelo de clasificación supervisado para señáles EEG que permita la comunicación en pacientes con neuropatía periférica.

Abstract

Peripheral neuropathies are conditions of the peripheral nervous system that cause a sensation of numbness, pain, hypersensitivity, etc. This causes the upper and lower limbs to suffer certain degrees of muscular atrophy. This leaves the patient almost completely paralyzed. In addition, in certain cases it can affect the frontal lobe where the control of expressive language is considered. To facilitate communication in patients with this type of motor and speech disorders, there are several assistive technologies, one of the most promising being the brain-computer interface (BCI) based on electroencephalograms (EEG), which allows direct interaction between brain activity and an electronic device. Therefore, in this integrative work, machine learning techniques were applied to create a model for classifying EEG signals based on steady-state visual evoked potentials (SSVEP) that correspond to signals from responses to visual stimuli emitted at a certain time. frequency, thus achieving a maximum of five communication patterns for patients. The work consists of the evaluation and implementation of the model, where the process of training and validating the model is based on metrics related to what the state of the art dictates. Resulting in a model capable of classifying visual stimuli with frequencies of 13, 15, 17 and 19 Hz. Finally, inferring the model with four basic communication commands for anyone who wishes to communicate.