BCI-Ageing: BCI tools for promoting active ageing
Brain Computer Interface for cognitive training and domotic assistance against the effects of ageing.
Project overview
The BCI-Ageing project is aimed at developing new tools that may help elderly people in their daily living. The main objective focuses on the development of two Brain Computer Interface (BCI) applications for elderly people. Firstly, a BCI application to train cognitive abilities in order to prevent ageing effects will be developed. Secondly, a BCI application to aid dependent elderly people to lead a healthy and comfortable life at home will be implemented. The opinion and real needs of elderly people will be taken into account to develop useful products.
A BCI is a communication system that monitors the brain cortical activity and translates specific signal features, which reflect the user’s intent, into commands that operate a device. In BCIs, the method most commonly used for monitoring brain activity is the electroencephalogram (EEG), since it is a non-invasive method, easy to use and that requires relatively simple and inexpensive equipment.
According to the World Health Organization (WHO), ageing is a progressive, generalised impairment of function resulting in a loss of adaptive response to stress and in a growing risk of age associated disease. A domotic BCI application would allow elderly people to interact with their usual environment and increase their personal autonomy. Moreover, the mental tasks used for some BCIs training could be really suitable for people in early stages of ageing, in order to prevent cognitive deterioration.
Strategy
A cognitive training application will be developed using a sensorimotor-based BCI. This application will use the µ (8-12 Hz) and β (16-24 Hz) sensorimotor rhythms. The mentioned rhythms present variations in EEG over motor cortex when a self-generated movement is performed and also when a subject observes the movement or imagines it. Thirteen channels (4 pre-frontal, 5 central and 4 central-parietal) will be studied during the BCI-Ageing project.
In order to suitably identify the users’ intent, the following methods for feature extraction will be studied:
For feature selection and classification methods, machine learning algorithms will be studied. These algorithms allow the automatic construction of category descriptions from a set of samples belonging to each of the considered categories, so that these descriptions produce the maximum possible discrimination between them. The descriptions are then used as predictive models to classify new samples as belonging to one or more of the existent categories. The following methods for the feature selection and classification stages will be studied:
Finally, the proposed cognitive training tasks implemented in the BCI application will be designed taking into account the therapists and elderly people from the CRE-DyD. Proposed cognitive training tasks will be divided into different subsets and difficulty levels, in order to assess progression of the users’ skills. Thus, the resultant application will allow elderly people to train their cognitive abilities by means of mental tasks. The graphical interface will be user-friendly and simple to avoid that users can get distracted by external stimuli.
On the other hand, an assistive domotic application will be developed using a P300-based BCI. The application will be aimed at allowing dependent people to control the devices usually available at home, satisfying their main needs. The application could control different electronic and domotic home devices and their main functionalities:
This application will use the P300 evoked potential elicited by an infrequent stimulus when interspersed with frequent stimuli. The BCI system will identify which option, among several shown at the computer screen, elicits a P300 potential when it is flashed. Thus, it will identify the user’s intent and it will perform the desired command. For that purpose, EEG signals will be recorded from 8 channels mainly over parietal and occipital cortex. These signals will be analysed using different algorithms to accurately detect the P300 potentials:
