Social spatial navigation assessment for early-stage Alzheimer's disease

点击翻译按钮获取中文摘要

Project Summary Alzheimer’s disease (AD) is a public health epidemic, laying an enormous mental, physical, and financial burden on individuals, families, and societies. Despite significant research efforts around the world, the specific link between cognitive / functional deficits and the underlying pathology remains unclear. Critical to solving this is the evaluation of novel cognitive tests within cohorts that have detailed biomarkers for AD pathophysiology. Widely used cognitive tests for AD focus on episodic memory deficits, but recent work led by us and others indicates that tests of spatial navigation and orientation may provide significantly more accurate detection of the earliest signatures of AD. Our hypothesis is that integrating tests of navigation and orientation (to people, places, and events) into one test will prove particularly sensitive. Currently no such test exists. Our aim is to develop and evaluate a novel ecologically valid cognitive test which probes both navigation and orientation in order to assess cognitively normal older adults (CN), amnestic mild cognitive impairment (MCI), and mild AD dementia participants from two ongoing imaging studies (R01 AG053184 and R01 AG067021) of the Principal Investigator (PI), Dr. Gad Marshall. Participants will undergo amyloid and tau positron emission tomography (PET) as part of those studies, which we will leverage in relating those biomarker findings to our novel cognitive test across the early AD continuum. Creating the new test will require combining the expertise of Co-PI Dr. Hugo Spiers (expert in spatial navigation) and Co-PI Dr. Shahar Arzy (expert in orientation). Dr. Spiers has recently developed a virtual reality (VR)-based assessment tool for smartphone and tablet devices that has tested 4.3 million participants on their navigation ability. While this enables an unparalleled opportunity for machine learning to detect subtle impairments, the task is not tailored to the individual’s personal world, which clinical experience as well as research from the Co-PI, Dr. Arzy, indicate to be critical in AD. We therefore propose to develop a novel patient-tailored digital personalized tool for the diagnosis and monitoring of early- stage AD. Building on our recent work, Google Street View (GSV) images will be used to display and enable navigation of the participant’s familiar environment. Overlaid text messages will pop up to provide information about the overriding cover-story and to test navigation and orientation. We will benchmark how hard each person’s environment is to navigate using reinforcement learning (RL) agents trained on the local street networks. Functional magnetic resonance imaging (fMRI) will be used to understand the brain networks engaged by the new task and allow a comparison with the pathological and clinical data collected by Dr. Marshall. Machine learning models will be used to detect subtle impairment in the individual participant level. Following Dr. Spiers’ success in mass appeal and the access to advanced machine learning analyses, the project will enable precision personalized diagnostics of people at the earliest symptomatic stages of AD, and later on, a digital tool to conveniently follow patients over time.