Researchers at Stony Brook University, in collaboration with The Chinese University of Hong Kong, have developed a new artificial intelligence framework aimed at improving the detection and monitoring of Alzheimer’s disease. Nearly seven million Americans over the age of 65 currently live with Alzheimer’s, and that number is expected to almost double by 2060, according to data from the Alzheimer’s Association.
The research team includes Shan Lin, associate professor of Electrical and Computer Engineering at Stony Brook University; PhD candidate Heming Fu; and Guoliang Xing from The Chinese University of Hong Kong. They created SHADE-AD (Synthesizing Human Activity Datasets Embedded with AD features), a generative AI system designed to produce synthetic data that closely reflects the motor behaviors specific to Alzheimer’s patients.
The SHADE-AD model identifies movements such as stooped posture, reliance on armrests when standing up, or slowed gait—subtle actions that can signal early onset of Alzheimer’s. By replicating these patterns, SHADE-AD aims to provide researchers and physicians with more targeted data for improving diagnosis and ongoing patient monitoring.
Unlike existing generative models that typically use datasets based on healthy individuals, SHADE-AD was trained specifically to include traits associated with Alzheimer’s. It produces three-dimensional “skeleton videos,” which are simplified representations preserving joint motion details. These synthetic datasets were validated against real-world patient data. According to the research team, the model accurately reproduced changes in speed, angle, and range of motion characteristic of Alzheimer’s patients compared to healthy older adults.
Findings from this project were presented at the 23rd ACM Conference on Embedded Networked Sensor Systems (SenSys 2025). Activity recognition systems trained using SHADE-AD outperformed those trained on traditional augmentation or general open datasets in identifying key actions like walking and standing up—movements often affected early in disease progression.
Shan Lin commented on the potential impact: “If we can provide tools that spot these changes before they become severe, patients will have more options, and families will have more time to plan,” he said.
Lin also linked this work to broader efforts supporting healthy aging: “Healthy aging isn’t only about treating illness, but also about creating systems that allow people to thrive as they grow older,” he said. “AI can be a powerful ally in that mission.”
