Artificial intelligence could change how older adults and people with dementia are cared for, but very few AI tools have made the leap from research into everyday medical practice. A new effort at Duke University is aiming to close that gap. The Duke Artificial Intelligence and Technology Collaboratory for Aging: Governance and Evaluation (Duke AITC-AIGE) is one of five centers funded by the National Institute on Aging's Artificial Intelligence and Technology Collaboratories program, and it will receive nearly $25 million over five years.
The money is meant to build shared infrastructure, fund research projects across the country, harmonize aging-related databases, and provide expertise to help promising tools reach real-world use. Juliessa Pavon, MD, an associate professor in the Division of Geriatrics and one of three Duke principal investigators, said Duke is the only center led from the Southeast, pointing to the university's strengths in geriatrics, dementia research, translational science and responsible AI oversight. Pavon described the central problem plainly: the field has no shortage of algorithms, but it lacks AI solutions that have been rigorously evaluated, safely implemented, continuously monitored, and shown to improve outcomes for older adults.
The source describes several areas where AI could help people living with dementia. Possible uses include detecting early cognitive decline by analyzing speech, behavior, and wearable device data, identifying risks such as falls or wandering, supporting medication optimization and deprescribing, and helping caregivers monitor systems and get personalized guidance. Pavon said researchers are already seeing promising work in these areas, and the center aims to determine which approaches actually improve outcomes and quality of life.
A central element of the initiative is the development of model cards, standardized summaries that would describe what an AI tool is designed to do, the data it was trained on, how it performed in testing, where it works well, and where it may be less reliable. The article compares them to nutrition labels on food. Model cards would also flag potential biases and appropriate uses so clinicians, health systems, researchers, and patients can judge whether a tool fits a given situation. Pavon likened this to the way clinicians weigh a medication's risks and benefits before prescribing it.
The center will also make its resources available beyond large academic institutions. A national pilot funding program is intended to give investigators, community researchers, smaller institutions, and even patients access to expertise in AI evaluation, data science, implementation, ethics, governance, and stakeholder engagement. A secure platform called NeuroAI is designed to combine medical records, brain images, laboratory results, and genetic information for studying aging and dementia. Rishi Kamaleswaran, PhD, who leads the Clinical Implementation and Data Science Core, said the goal is to democratize AI expertise built at major academic centers. The stated aim over the long term is technologies that help older adults stay independent, support caregivers, and improve care delivery.