UCLA Health will lead a national programme to test and evaluate artificial intelligence (AI) tools for people living with Alzheimer’s disease and other dementias after receiving a five‑year, US$25 million grant from the US National Institute on Aging (NIA), according to the university.
National validation effort for AI in dementia care
The initiative, named Accelerating Platforms for Evaluating and Executing AI Lifecycles in Alzheimer’s Disease and Related Dementias (APEX‑ADRD), will support a coordinated, nationwide approach to assess how AI systems perform in real‑world clinical settings. UCLA Health said the programme aims to bring “rigour to every stage of an AI tool’s journey” — including the data used to train tools, their development and their clinical performance.
APEX‑ADRD will be a collaborative effort that includes other major health institutions such as the Mayo Clinic and the University of Wisconsin, according to the announcement.
“As we continue to adapt to the rapidly evolving AI landscape, this new venture represents a significant step forward for UCLA Health as a centre of excellence in the evaluation of AI tools to support clinicians,” said Johnese Spisso, president of UCLA Health and chief executive officer of the UCLA Hospital System, as reported by UCLA Health.
UCLA reported that the grant will fund activities to validate technologies designed for the diagnosis and treatment of patients living with Alzheimer’s and related dementias, with the goal of helping clinicians identify cognitive decline earlier, personalise care plans, support caregivers and reduce administrative burdens.
Why validation matters
Proponents of the project say systematic evaluation is needed because AI tools can vary widely in the data they are trained on, the populations they were developed for and the settings in which they are deployed. Dr John N. Mafi, associate professor of medicine and co‑director of UCLA Health’s Innovations and Outcomes Validation of AI (INOVAi) Center, said the programme will support teams across the country at every stage of development and evaluation.
“Older adults living with dementia have the most to gain and the most to lose,” Dr Mafi said, according to UCLA Health. “Our programme will bring rigour to every stage of an AI tool’s journey – the data it learns from, how it is built and how it performs in real‑world care.”
UCLA noted that more than 7.2 million Americans are living with Alzheimer’s disease, a figure expected to rise in coming decades. The institution said APEX‑ADRD could help standardise evaluation methods so that health systems and researchers can compare and understand the benefits and limitations of emerging AI technologies.
Planned activities and potential benefits
The announcement states the programme will fund and support teams across the country at every step, although details about the specific projects to be funded, timelines for deployment and metrics for success were not included in the press material.
UCLA highlighted several areas where validated AI tools might be useful:
- Earlier identification of cognitive decline and more accurate diagnosis.
- Personalised care planning tailored to a patient’s needs and circumstances.
- Support for family carers through decision aids or monitoring tools.
- Reduction of administrative burden on clinicians through workflow automation.
The university framed APEX‑ADRD as a national resource that will both fund and guide evaluation work, with a specific emphasis on ensuring that technologies demonstrate benefit across diverse populations in California and beyond.
Context and caution
Interest in clinical AI has surged amid promises of improved diagnosis and efficiencies, but researchers and regulators have cautioned that tools trained on limited or unrepresentative data can amplify bias and produce unsafe recommendations. Systematic, independent validation is a recurring recommendation from clinical and regulatory experts.
While the grant and the creation of APEX‑ADRD signal an investment in those validation safeguards, the ultimate impact will depend on how the programme operationalises its evaluation standards, the transparency of its findings and how quickly validated tools are adopted in clinical practice.
Patients and caregivers considering changes to diagnosis or treatment should consult their clinicians or local clinics rather than rely on AI tools directly; validated AI is intended to support, not replace, clinical judgement.
