Biosensors for the detection of thiol-containing biomacromolecules via cationic aromatic compounds

Investigator: Reza Abiri, University of Rhode Island 
Mentor: Walter Besio, University of Rhode Island

Theme: Neuroscience
Title: Biosensors for the detection of thiol-containing biomacromolecules via cationic aromatic compounds
Award: Early Career Development (2022-2024)

Abstract: The ability to focus attention and encode the information are among the brain’s most important cognitive and perceptive functions. Different neurological diseases such as attention deficit hyperactivity disorder (ADHD) and Alzheimer’s disease (AD) can result in severe attentional deficit. Recently, neurofeedback-based functional magnetic resonance imaging (fMRI) training showed promising results in cognitive rehabilitation and attention remedy. However, customization of such training protocol per individual using traditional neurofeedback methods is often difficult and labor- and time-intensive especially when dealing with patients with cognitive impairment. In this project, real-time electroencephalography (EEG)-based brain-computer interface (BCI) technology is leveraged to accelerate the development of neurofeedback-based attention training protocols through clinical settings and public usage. To this end, Aim 1 of this study focuses on developing novel closed-loop adaptative algorithms for neurofeedback-based attention training.

In this mechanism, the neural decoder is being re-fitted continuously to previously recorded sample points and finally co-adapts to the human subject’s neural activity for the highest reward in feedback. In other words, this adaptation protocol helps human subjects to stabilize their neural strategies in the designed attention task. The stabilized neural maps are the attention neuromarkers recorded from whole brain of human subjects. The anticipation is such neuromarkers from a population study carry more similar features/activation compared to traditional neuromarker extracted using nonadaptive neurofeedback. In Aim 2, powerful dimensionality reduction methods from computational neuroscience and deep learning are adopted to find latent neural space within and across subjects in Aim 1. The existence and temporal convergence of latent neural dynamics across subjects will be investigated. A generic neural attention model will be developed using shared variance and latent space. This generic model in combination with transfer learning theory can help to extract an attention model for a new human/patient subject with limited data access. The attention model will be tested with a leave one subject out cross-validation technique.

Relevance: The proposed study accelerates the translational aspect of neurofeedback-based attention training using novel technologies, algorithms, and neuromarkers. The presented brain training intervention and approaches will not only be applicable to attention remedy in patients with attention deficits but also can be applied to treatment of other neurocognitive (mental) disorders such as anxiety, depression, and addiction.