Randomized trial demonstrates a camera-based system for detecting getting-up postures in elderly, indicating potential for enhanced safety and reduced staff workload.
Key Points
To develop a camera-based system that detects elderly getting-up postures to prevent bed falls.
Utilized a YOLOv11 deep learning model for posture detection
Collected data in laboratory and hospital settings with 19 and 5 participants respectively
Developed a multilabel dataset for training under varied conditions
Achieved average accuracy of 99.4%, sensitivity of 98.8%, and specificity of 99.7%
Successfully detected getting-up posture and tracked head movement beyond bed boundaries
System issues alerts and estimates visible heads effectively