Al-Supported In-Home Brain Assessments for Older Adults and Persons with Alzheimer's Disease

University Of Massachusetts Amherst, Amherst MA

Investigators

Linked publications, trials & patents

Abstract

Project Summary More than 5 million Americans are living with AD—the sixth leading cause of death in the U.S. and the only leading cause of death that cannot be prevented, cured, or substantially slowed. Early and accurate diagnosis can save up to $7.9 trillion in medical and care costs and is therefore a critical need. However, going to the hospital for health evaluations can be onerous for everyone, particularly so for older adults with limited mobility, vision and/or cognition, as well as variable access to transportation and home caregivers. Home-based assessment could be a solution, but existing assessment tools for AD—including PET, MRI, and biomarkers in cerebrospinal fluid or plasma—are cumbersome, expensive, or invasive, making them ill-suited for home use or even regular screening purposes. An easy-to-use, home-based screening test with good sensitivity and specificity for identifying AD would thus be of tremendous value and fill an unmet clinical need. No such technology currently exists. Over the past 15 years, our group has been developing battery-powered wearable neurovascular and physiological monitoring technologies to facilitate applications ranging from seamless perioperative monitoring to brain assessment in spaceflight. These devices have the potential to act as a ‘mobile clinic’ to support brain-based assessments at home, and have been utilized in numerous remote and self-deployed settings. They have not, however, been assessed for their feasibility for use by older adults, AD patients. In this project, we propose to achieve the following specific aims. Aim 1: Generate an adapted version of NINscan so older adults or AD patients can collect high quality brain and physiological data at home. Aim 2: Generate and share a database from home data collections during rest and cognitive tasks in older adults and AD patients recruited from the MassAITC-affiliated MADRC Core. Aim 3: Use transformer- based artificial intelligence (AI) models to (i) identify individuals with suspected AD, (ii) predict cognitive testing scores, and (iii) predict plasma tau levels, all using the neural and vascular biomarkers collected during self- (or caregiver-) deployed recordings at home. The results of this project will help make brain-focused AD-relevant testing easier for both patients and clinicians, and could potentially suggest a more cost-effective and sustainable future health care model for AD.