An AI-assisted screening platform within a multivariate framework for biomarkers of mild cognitive impairment due to Alzheimer's disease

点击翻译按钮获取中文摘要

PROJECT SUMMARY In the past decade, the rate of deaths from Alzheimer's disease (AD) and other dementias escalated more than twice the rate of deaths from heart disease. Unfortunately, there is a lack of low-cost and non-invasive diagnostic instruments to accurately identify individuals at risk of AD and ADRD. Advanced non-invasive imaging shows that retinal neurodegeneration and visual deficits occur long before the cognitive decline in AD and ADRD. This fact raises the possibility of identifying mechanisms that drive retinal pathology in AD/ADRD that could help develop effective diagnostics tools and therapies that target early disease. The well-characterized organization of the retina, with powerful non-invasive imaging and electrophysiology techniques to monitor retinal function, make it an optimal surrogate to study early CNS pathology. The brain shares many similarities with the retina. This suggests that the retina, a more accessible organ than the cortex, may provide a viable brain biomarker for testing diagnostics tools and therapies that target early disease and prevention. Notably, we happen to live in a non-linear world surrounded by objects and processes with the property of fractality and non-linearity. For example, the deficit of fractal complexity (i.e., fractality) of environmental effects can lead to fractal complexity distortion in the brain's visual pathways and abnormalities of development or aging. Particularly, non-linear dynamics of physiological processes involved in neurodegenerative disorders have a strong base of evidence, which is seen in the impaired fractal regulation of rhythmic activity in aged and diseased brains. Our multivariate biomarker methodology relies on the fractal complexity of the retinal vasculature as a potential biomarker. However, the fractality of the time-varying electroretinogram (ERG) signal that arises from different retina layers is not yet explored. Therefore, we aim to take advantage of the current electrophysiological measurements acquired in the parent grant to investigate the distortion of fractal complexity in ERG signals correlated to AD pathology as a possible means to obtain a more comprehensive assessment for the early detection of MCI due to AD. In this project, we will further innovate our multivariate biomarker methodology by investigating the fractality of ERG signals. This investigation would make our novel method a more robust tool by incorporating the combined fractality of the retinal function (ERG signals) and structure (retinal vasculature), which can shed new light on early pathogenic mechanisms that compromise retinal and brain function much before the onset of detectable dementia. To this end, we will investigate the distortion of fractality in ERG signals and explore the discrimination power of ERG's fractality measurements between groups with the receiver operating characteristic curve, sensitivity, and specificity metrics. We will use the Youden index and the area under the curve will be calculated for the ERG device calculated features. This project may enable a more comprehensive assessment of aging on ocular and cerebral function at the early stage of cognitive impairment by identifying the most initial signs of complications in the eye and brain using relevant multimodal measures of ocular abnormalities.