Artificial Intelligence for Early Detection of Alzheimer's Disease: A Comparative Review of Speech, Neuroimaging, and EEG Biomarkers

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Alzheimer's disease (AD) is a chronic neurodegenerative disorder and the most common cause of dementia globally. Accurate and early diagnosis of AD is essential to enhance patient care and management, facilitate recruitment for clinical trials, and provide access to novel treatments at a stage when intervention may have greater potential benefit. However, the early detection of AD before obvious clinical symptoms have emerged is challenging. Firstly, there is no unique, widely available screening test for early diagnosis of AD. Secondly, AD progresses through a series of brain and cognitive alterations that resemble those that occur during the typical aging process, making detection of the disease in its early stages difficult since these subtle signs and symptoms can often be mistaken for aging. This overlap between normal aging and early AD-related changes further complicates efforts to identify individuals at risk before substantial cognitive decline occurs. While a combination of neuropsychological assessment, neuroimaging, and analysis of cerebrospinal fluid (CSF) biomarkers can facilitate the diagnosis of AD once symptoms have manifested, these diagnostic tools are not ideal for widespread pre-symptomatic screening due to their high costs, invasiveness, and/or limited availability. In recent years, artificial intelligence (AI) techniques have been increasingly employed to pre-symptomatically identify subtle changes in speech and language, neuroimaging, and electrophysiological activity including electroencephalography (EEG) and quantitative EEG (qEEG) that are predictive of AD onset. This comparative review discusses recent original studies that have investigated the potential of AI-powered approaches in these three modalities for early detection of AD and discusses the challenges that need to be overcome for these techniques to translate into routine clinical practice.