23L004 A System for Dementia Risk Prediction from Movement Data

With the availability of wearable devices, it is easy to capture people"s physical activity. Our team has designed a novel deep learning model which is trained from movement data of healthy subjects and subjects with Alzheimer"s Disease. It uses a time-aware attention mechanism to model the effect of circadian rhythms. Our model takes the movement data of a subject as input, and generates meaningful analysis and visualization of his/her daily behavior patterns (e.g., sleeping, exercising, light activity, etc.), and provides a graphical indication of potential dementia risk.