started · updated
Weizmann Institute researchers use AI to link movement patterns to health conditions
Researchers at the Weizmann Institute of Science have developed an artificial intelligence model capable of identifying health issues through body movement analysis. According to a study published in Nature Medicine, the system analyzed 3,328 adults performing five specific tasks: walking on a treadmill at 3 km/h, walking at a preferred pace, walking in place, rising from a chair, and maintaining balance while standing.
Using a depth camera to record three-dimensional joint movements, the AI model identified associations with 1,887 out of 3,210 medical variables examined. Notably, the technology could predict liver stiffness with accuracy comparable to blood tests. It also identified patterns related to sleep apnea, insomnia, and sleep duration. Furthermore, combining movement data with traditional physical indicators like age, height, and body fat improved predictions for conditions such as diabetes and depression.
While scientists hope this non-invasive technology will complement medical checkups, the study was cross-sectional, meaning it cannot yet establish cause-and-effect relationships or determine if health changes precede movement alterations.