What if your smartwatch could detect changes in your cognitive health?
Early detection of cognitive decline can be challenging because Alzheimer’s disease can affect memory, language, reasoning, behavior and everyday functioning. Current evaluations can combine cognitive assessments with medical history, laboratory tests and brain imaging, while researchers continue to explore ways to identify and track subtle changes earlier, including through digital and behavioral measures.
IBM is proposing an AI system that would use data from wearable and other connected devices to monitor changes in a user’s cognitive state. The system would build a personalized user profile, compare it with historical data and use an AI model to generate long-term and real-time cognitive state scores for assessing Alzheimer’s risk.
How it works
The proposed system receives data from wearable devices and Internet of Things (IoT) networks to monitor changes in a user’s cognitive state. It creates a personalized user profile based on the collected data and uses the profile together with historical user profiles to train an AI model. A random forest algorithm can then analyze this information to generate cognitive state scores (CSS).
The system generates two types of CSS. The long-term CSS is intended to capture changes over a defined period, while the real-time CSS reflects the user’s current cognitive state. Together, these scores can be used to calculate an Alzheimer’s risk score and compare it with a predetermined threshold.
If the resulting risk assessment reaches a specified threshold, the proposed system can trigger a response, such as sending a notification to a caregiver. It can also dynamically adjust a geofence based on the user’s real-time cognitive state, potentially providing an additional layer of protection when the system detects a relevant change.
Why it matters
The proposed system builds on IBM’s broader research into artificial intelligence and Alzheimer’s disease assessment. IBM Research and Pfizer previously developed an AI model that used short speech samples to help predict the eventual onset of Alzheimer’s disease in cognitively healthy individuals. IBM has also explored AI-based cognitive assessment through speech analysis and machine-learning models for detecting early cognitive impairment.
The application takes a different approach by combining data from connected devices instead of relying on a single assessment. By comparing current information with personalized and historical user profiles, the system is designed to identify changes in an individual’s cognitive state and provide risk-related notifications and protective measures.
As researchers explore digital and behavioral measures for Alzheimer’s detection and monitoring, IBM’s proposed technology shows how connected devices could support more continuous and personalized cognitive health assessment.
U.S. Pat. App. Pub. No. 2026/0279582, titled “Alzheimer’s Detection and Patient Protection,” was filed on March 13, 2025, and was published on September 19, 2026. The listed inventors are Yu Zhu, Su Liu, Peng Hui Jiang, Guang Han Sui, Jun Su, and Jun Feng Duan.