INNOVATIVE EEG BIOMARKERS AND AI ALGORITHMS FOR THE EVALUATION OF PHYSIOLOGICAL AND PATHOLOGICAL AGING

Aging is a biologically natural process that changes the structure and functions gradually at different levels, from the cells to the large-scale brain networks. Usually, physiological aging is associated with mild cognitive change; however, in a significant number of people, the process can become pathological and eventually result in Mild Cognitive Impairment (MCI) and Alzheimer's disease (AD). Differentiating healthy from pathologic aging is still a hard scientific and clinical problem, especially in the preclinical and prodromal phases of neurodegenerative diseases. One method to study age-related brain changes is the electroencephalography (EEG), a non-invasive technique with a very high temporal resolution. Initial EEG studies have already contributed significantly to the understanding of age-related changes in brain oscillatory dynamics, but the combination of next-generation brain analytic methods, such as graph theory and entropy analysis, has the potential to reveal new biomarkers that can indicate not only physiological adaptations but also pathological deviations. Besides that, the new implementation of machine learning and artificial intelligence (AI) in the neurophysiological field represents a powerful tool for predictive and individualization purposes. The current PhD project, titled "Identification of innovative EEG biomarkers and AI algorithms for physiological and pathological aging evaluation", aims to identify novel EEG biomarkers and implement AI models to distinguish between age groups and cognitive states. Specifically, the project is meant to achieve three main goals: (i) through connectivity and entropy analysis to validate novel EEG features that show statistically significant differences not only between young, adult, and elderly subjects but also between healthy elderly controls, MCI subjects, and AD patients; (ii) to implement AI algorithms that can track brain age changes among healthy people and differentiate healthy elderly people from pathological group, with the ultimate goal of facilitating early detection, prognostic assessment, and treatment planning. 120 healthy volunteers equally spread across three age groups (young, adult, and old) and 40 MCI as well as 40 AD patients were enrolled. Additionally, data from databases were included, for a total of 109 healthy elderly subjects, 75 MCI subjects, and 85 AD patients. Resting-state EEG in eyes-closed condition will be recorded from all participants. Healthy elderly, MCI subjects, and AD patients will be evaluated through a battery of neuropsychological tests. Functional connectivity will be computed along with graph theoretic measures, and signal complexity will be described by Approximate Entropy. Additionally, to both brain age estimation and physiological 3 versus pathological aging classification, optimization and feature selection methods will be used to pinpoint the most discriminative EEG-derived predictors. These analyses allow to significantly raise the classification's precision and reliability and reveal the neurophysiological features with the most significant diagnostic and prognostic value. The research project would be a significant contribution to the emerging field of computational neurophysiology by facilitating the further application of EEG biomarkers and AI models. It would potentially lead to the implementation of robust biomarkers of aging which could be used for early detection of neurodegenerative processes, prognostic refinement, and, ultimately, the advancement of personalized therapy in the domain of age-related cognitive decline.