Brett Bormann

The ability to comprehend speech in noisy environments is extremely variable, even for people with clinically normal hearing. My research uses machine learning to better understand the neural mechanisms of attention in the ascending auditory pathway. This analysis is a part of a larger project working to design more sensitive brain-based hearing loss diagnostic tools.

The majority of EEG research of auditory attention has primarily focused on the high-level processing areas in the cortex of the brain but does not investigate the attentional modulations of neural activity during the earliest stages of auditory processing. This project aims to better understand the functional connectivity of the brainstem, thalamus, and cortex while participants listen to our novel continuous chirped speech (cheech) stimulus. To understand the interconnected pattern of activity across these brain areas, we used partial least squares (PLS) which is a machine learning method that can identify nuanced correlations that would not be detected with other regression modeling techniques. These methods enabled us to identify the functional connectivity from cochlea to cortex. We were also able to show that the role of attention is recruited far earlier during auditory processing than previously believed. We aim to further utilize these methods to better understand these relationships while also using it to train machine learning classifiers for our goal of creating brain-based hearing loss diagnostic tools.