A machine-learning method isolating changes in wrist kinematics that identify age-related changes in arm movement

Area of Science:

• Gerontology
• Biomedical Engineering
• Movement Science

Background:

• Normal aging is associated with physiological tremors and reduced hand movement speed, impacting daily life.
• Quantifying age-related motor changes is crucial for understanding functional decline.

Purpose of the Study:

• To detect and identify age-related changes in wrist kinematics and response latency using wearable sensors.
• To differentiate kinematic patterns between young and older adults during specific tasks.

Main Methods:

• Utilized lightweight, non-invasive wearable inertial measurement units (IMUs) on the wrists of young and older adults.
• Collected kinematic data during postural and pronation-supination tasks.
• Performed frequency analysis and angular velocity analysis on sensor data.

Main Results:

• Identified 5 kinematic variables distinguishing older from younger adults in a postural task (9-13 Hz range), achieving 0.86 AUROC and 89% accuracy.
• Detected a 71 ms delay in arm movement initiation for older adults during a pronation-supination task.
• Demonstrated reliability, sensitivity, and accuracy of kinematic analysis with commercial sensors.

Conclusions:

• Wearable sensor-based kinematic analysis reliably detects age-related increases in physiological tremor and motor slowing.
• This non-invasive method accurately quantifies age-associated motor function changes.
• Further research is needed to assess its efficacy in distinguishing physiological tremors from pathological tremors in neurological diseases.