P3‐177: Combining classical conditioning and brain‐state classification for the development of a brain‐computer interface (BCI) for Alzheimer's patients

Abstract

Background Alzheimer's Disease (AD) patients who have lost the ability to communicate verbally may benefit from a Brain-Computer Interface (BCI), which could allow them to convey basic thoughts and emotions, e.g. by associating affirmative and non-affirmative thinking to a positive and a negative emotion, respectively. One possibility to develop a BCI that can be used with AD patients is to modulate brain responses with a semantic classical conditioning paradigm, in order to associate congruent and incongruent word pairs with emotionally positive and negative stimuli, facilitating the discrimination between “affirmative” and “non-affirmative” thinking. Using a classical conditioning paradigm is convenient, since it does not require AD patients to put active effort in learning how the BCI works. Methods In our classical conditioning paradigm, already tested on 11 healthy subjects, the unconditioned stimuli consist of a positive and negative emotional sound (a baby laugh and a scream, respectively). The conditioned stimuli, presented aurally, are congruent (e.g. 'fruit-apple') and incongruent (e.g. 'fruit-dog') word pairs. During the conditioning acquisition, congruent and incongruent word pairs are associated to the baby laugh and the scream respectively. Functional magnetic resonance imaging (fMRI) will be performed on AD patients (Mini Mental State Examination score: 18-24; intact auditory system) on a 3T scanner. To classify the signals corresponding to congruent and incongruent word pairs, both univariate (General Linear Model) and multivariate (Support Vector Machine, SVM) analyses will be performed. Results Both univariate and multivatiate analyses on healthy subjects confirmed the possibility to discriminate between affirmative and non-affirmative responses. We hypothesize that a similar differentiation may be found in AD patients, since classical conditioning was demonstrated to be possible, at least to some extent, in these patients.The study with AD patients is ongoing, and preliminary results will be presented during the conference. Conclusions The encouraging results obtained with healthy subjects show that basic affirmative/non affirmative thinking discrimination is possible within an fMRI-BCI setting. The development of a BCI for AD patients could be an important step for allowing not only basic communication, but possibly lending a path for rehabilitation and diagnosis.