Trajectory and Sway Prediction Towards Fall Prevention.

Area of Science:

• Biomechanics
• Gerontology
• Computer Vision

Background:

• Falls are a major cause of injury and death, especially in older adults.
• Imbalance leading to falls stems from internal factors (illness) or external perturbations (active or passive).
• Understanding torso movement is crucial for fall risk assessment.

Purpose of the Study:

• To propose a novel metric for monitoring torso sway.
• To correlate torso sway with active and passive perturbations.
• To develop a predictive model for torso sway and fall risk.

Main Methods:

• Development of a torso sway monitoring metric.
• Correlation analysis between torso sway and perturbation types.
• Machine learning model using past trajectory, motion, and scene data for prediction.

Main Results:

• Large changes in torso sway are strongly correlated with active perturbations.
• The model can reasonably predict future torso sway and path.
• Visual cues from the surrounding scene improve prediction accuracy.

Conclusions:

• Torso sway is a reliable indicator of perturbations leading to falls.
• The proposed metric and predictive model show promise for fall prevention applications.
• Integrating visual scene information enhances fall risk prediction in older adults.