Vanderbilt develops AI tool to speed Alzheimer's patient referrals and treatment

A patient sits in a primary care office with a diagnosis of early cognitive decline. The doctor knows there are new drugs that can slow the progression—monoclonal antibodies that work in the earliest stages of Alzheimer's disease, when amyloid proteins are accumulating in the brain but symptoms are still mild. But between that moment and the first infusion lies a gauntlet: a referral to neurology, specialist evaluation, brain imaging, insurance approval. Weeks pass. By the time the patient finally receives treatment, the disease has continued its quiet work.

Vanderbilt Health researchers have decided to compress that timeline. With a $600,000 grant from Eli Lilly and Company, they are building an artificial intelligence tool embedded directly into the electronic health record system—the same software clinicians use every day to document patient care. The tool's job is simple in concept but consequential in practice: when a primary care doctor or geriatrician enters a referral for cognitive concerns, the AI agent will immediately summarize the relevant information from the patient's chart, flag the details that are commonly missing and that slow down specialist evaluation, and recommend whether the case should be routed as urgent or standard priority. Clinicians retain full authority to accept, edit, or override every suggestion the system makes.

You Chen, an associate professor of biomedical informatics, is leading the 18-month project alongside neurologist Amalia Peterson and geriatrician Sean Huang. The team's first task is archaeological: they will reconstruct the care timelines of more than 5,300 Vanderbilt patients who have already gone through this pathway, using AI to identify exactly where delays accumulate and why. Once they understand the friction points—whether it's missing imaging results, incomplete cognitive testing, or unclear insurance status—they can design the triage agent to address them proactively.

"By the time a patient reaches our clinic, the clock has often been running for weeks," Peterson said. The new tool is designed to ensure that when a referral arrives at neurology, the specialists already have what they need to move forward without delay. It is not meant to replace clinical judgment but to remove the administrative and informational obstacles that keep patients waiting.

The project is designed from the start to be transferable to other health systems. If the pilot succeeds—measured by how much time the system saves between diagnosis and first infusion—the team envisions expanding the approach across the entire Alzheimer's care pathway. A second AI agent could analyze MRI images and generate safety reports for radiologists to review. A third could assemble insurance authorization packets and track the approval process, another common bottleneck. The vision is a fully streamlined handoff from primary care to specialty care, with artificial intelligence handling the coordination work that currently consumes weeks.

Chen noted that Vanderbilt is already leading a separate $1 million Lilly-funded project addressing gaps in obesity care. These grants, he said, reflect the institution's commitment to using AI not as a replacement for clinical care but as a tool to eliminate the delays and friction that exist between diagnosis and treatment. For Alzheimer's patients, where time is literally a measure of disease progression, that difference could mean the gap between early intervention and advanced decline.